91577 DISCUSSION PAPER NO. 1419 Rwanda Social Safety Net Assessment Alex Kamurase, Emily Wylde, Stephen Hitimana and Anka Kitunzi Africa Social Safety Net and Social Protection Assessment Series July 2012 RWANDA Social Safety Net Assessment Alex Kamurase, Emily Wylde, Stephen Hitimana and Anka Kitunzi July 2012 Africa Social Safety Net and Social Protection Assessment Series Until recently, most countries in Africa implemented safety nets and social protection programs only on an ad hoc basis. In the wake of the global economic, food and fuel price crises starting in 2008, however, policymakers in Africa began to increasingly view safety nets as core instruments for reducing poverty, addressing inequality, and helping poor and vulnerable households to manage risk more effectively. During FY2009-2013, to support governments in their quest to understand better how to improve the efficiency and effectiveness of safety nets in their countries, the World Bank’s Africa Region undertook social safety net or social protection assessments in a number of countries in Sub-Saharan Africa. By 2014 assessments have been completed or are under preparation for over 25 countries in sub-Saharan Africa. These assessments analyze the status of social protection programs and safety nets, their strengths and weaknesses and identify areas for improvement, all with the aim of helping governments and donors to strengthen African safety net systems and social protection programs to protect and promote poor and vulnerable people. They were all carried-out with the explicit aim of informing governments’ social protection policies and programs. With the results of analytical work like these assessments and other types of support, safety nets and social protection programs are rapidly changing across Africa. For a cross-country regional review, please see "Reducing Poverty and Investing in People: The New Role of Safety Nets in Africa," which pulls together the findings and lessons learned from these assessments and other recent studies of safety net programs in Africa. Abstract Rwanda has experienced remarkable reductions in poverty from 59 percent in 2000/01 to 45 percent in 2010/11 (NISR, 2012). Social protection sector has equally evolved recently, both in terms of extending coverage of new programs and developing core functions on policy and strategy. The sector has matured from fragmented and largely off-budget donor programs to an increasingly government-owned and coherent system of interventions. But many Rwandans remain in extreme deprivation and are often stuck in chronic poverty. The sector thus needs to evolve further by deepening systems and reforms to the implementation level, increasing current low coverage of the poor by the core social safety nets programs and adequately responding to the changing needs of poverty and vulnerability. The government remains committed to pro-poor reforms and ensuring inclusive growth is sustained and enhanced. JEL Classification: I32, I38, J32, H53 Keywords: social protection, systems, safety nets, social assistance, welfare, administration, public policy, public sector reform, developing countries ABBREVIATION AND ACRONYM DAD Development Assistance Database DFID Department for International Development (UK) EDPRS Economic Development and Poverty Reduction Strategy EICV Enquête Intégrale de Conditions de Vie de ménages (household survey) IMF International Monetary Fund MIGEPROF Ministry of Gender and Family Promotion MINALOC Ministry of Local Government MINECOFIN Ministry of Finance and Economic Planning MINISANTE Ministry of Health MIS Management Information System MTEF Medium-Term Expenditure Framework PEFA Public Expenditure and Financial Accountability PER Public Expenditure Review PRIMATURE Office of the Prime Minister S-JAF Secteur Joint Action Forum SP Social Protection SPWG Social Protection Working Group SWAp Sector-Wide Approach UNICEF United Nations Children’s Fund VUP Vision 2020 Umurenge Programme WFP World Food Programme i ACKNOWLEDGEMENT This document is a product of collaboration between MINALOC and the World Bank. The focal point for this assessment in the Ministry of Local Government is Ms. Francine Tumushime, the Director General of Social Affairs and Community Development. Mr. Egide Rugamba, the Director General of Planning in the Ministry of Local Government and Mr. Justine Gatsinzi, the Deputy Director General for the Rwanda Local Development Support Fund also provided invaluable guidance. The assessment was done by a World Bank team led by Alex Kamurase (Senior Social Protection Specialist). The consultancy team included Emily Wylde (Social Policy Consultant and Main Author), with support from Stephen Hitimana (PFM consultant) and Anka Kitunzi (Fiscal Decentralization consultant). The team benefitted from the guidance and advice of Yasser El-Gammal (Cluster Leader for Rwanda), Lynne Sherburne- Benz (Sector Manager for Social Protection in Africa) and Mimi Ladipo (Country Manager for Rwanda). The report benefitted from peer review comments from John Elder (Lead Economist), Ruslan Yemtsov (Lead Economist) and Afrah A. Al-Ahmadi (Senior Human Development Specialist). The team are grateful to many individuals for their helpful inputs, including in particular Elias Baingana, Director General for National Budget; Leonard Rugwabiza, Director General for National Development Planning and Christophe Nsengiyaremye, Director of Inter-governmental Transfers Unit in MINECOFIN; Mayors, Vice-Mayors in Charge of Social Development, and Executive Secretaries in Bugesera, Kicukiro, Musanze, and Rwamagana who were generous with both their time and their insight; Cris Ayebare, EDPRS Social Protection Facilitator in charge of M & E in MINALOC; as well as other development partners and stakeholders from the Social Protection Working Group. ii Table of Contents ABBREVIATION AND ACRONYM ................................................................................... I ACKNOWLEDGEMENT ................................................................................................. II EXECUTIVE SUMMARY ............................................................................................... VI 1. INTRODUCTION ................................................................................................... 1 1.1 Background and context ..........................................................................................1 1.2 Objectives of the Assessment ..................................................................................2 1.3 Note on methodology .............................................................................................. 5 1.4 Structure of the report............................................................................................. 8 2. POVERTY, RISK, AND VULNERABILITY IN RWANDA ............................................... 9 2.1 Trends in poverty over time ..................................................................................11 2.2 Understanding poverty: who are the poor? .......................................................... 16 2.3 Risk and vulnerability ............................................................................................. 24 2.4 Conclusions on poverty and vulnerability ............................................................. 26 3. CURRENT LANDSCAPE: OVERVIEW OF RWANDA’S SOCIAL PROTECTION SECTOR 28 3.1 The evolution of the social protection sector in Rwanda ......................................28 3.2 Institutional arrangements ....................................................................................36 3.3 Expenditure patterns ............................................................................................. 37 3.4 Trends in beneficiaries ........................................................................................... 45 3.5 Summary of key existing issues and challenges and the implications for the assessment ......................................................................................................................... 47 4. SYSTEMS REVIEW: FROM PROGRAMME DESIGN TO IMPLEMENTATION ............. 53 4.1 Social Assistance: system of cash transfers ........................................................... 54 4.2 Social insurance: extension of contributory social security and labour standards ... ................................................................................................................................ 71 4.3 Social development programmes ..........................................................................75 4.4 Complementary services .......................................................................................80 4.5 Achievements and gaps ......................................................................................... 80 4.6 Overlaps and duplications .....................................................................................82 5. TOWARDS THE NEXT PHASE OF THE SOCIAL PROTECTION SECTOR: HURDLES TO OVERCOME AND A VISION FOR THE FUTURE ............................................................. 85 5.1 Harmonisation and streamlining of programmes .................................................85 5.2 Decentralisation and implications for social protection........................................94 5.3 Institutional arrangements for policy and oversight ...........................................107 5.4 Donor coordination: establishment of a SWAp ...................................................112 6. CONCLUSIONS AND RECOMMENDATIONS ....................................................... 116 6.1 Appropriateness and relevance of social protection programmes .....................116 6.2 Issues related to the effectiveness of specific programmes ...............................118 6.3 Higher-level strategic issues for the sector .........................................................120 6.4 Recommendations ...............................................................................................122 BIBLIOGRAPHY........................................................................................................ 127 ANNEX A CONCEPT NOTE .................................................................................... 130 ANNEX B UBUDEHE MATCHING EXERCISE METHODOLOGY NOTE......................... 139 iii List of Tables Table 1: Education indicators, EICV2 (2005/6) and EICV3 (2010/11) ............................... 14 Table 2: Childhood mortality rates ...................................................................................15 Table 3: Poverty incidence by main source of income .....................................................19 Table 4: Lighting source by consumption quintile, national .............................................21 Table 5: Household durable ownership distribution across quintiles .............................. 21 Table 6: Distribution of livestock ownership across consumption quintiles ....................22 Table 7: Correlates of poverty – regression results .......................................................... 23 Table 8: Health insurance coverage and physical access to health centres, EICV2 (2005/6) and EICV3 (2010/11) ........................................................................................... 25 Table 9: Potential future social transfer programmes as part of the social protection floor ....................................................................................................................................36 Table 10: Expenditure on the main cash transfer programmes, 2008-2010/11 actual, RwF million .........................................................................................................................40 Table 11: Expenditure on social protection, actual, 2004 to 2010/11, RwF million (nominal) ............................................................................................................................41 Table 12: Summary of shelter and income-generating activities .....................................83 Table 13: Lessons learned in Mozambique and implications for Rwanda .....................115 Table 14: FARG Direct Support beneficiary list matching with Ubudehe database by district ..............................................................................................................................140 Table 15: Age distribution of FARG beneficiaries ...........................................................141 List of Figures Figure 1: Real GDP growth per annum in Rwanda compared to EAC and SSA Average ..10 Figure 2: Poverty and extreme poverty incidence, 2000/01 to 2009/10 ......................... 12 Figure 3: Incidence of poverty 2010/11 by district ........................................................... 12 Figure 4: Growth incidence curves, 2001/2 – 2005/6 and 2005/6 – 2010/11 .................13 Figure 5: Under-five mortality by socio-economic characteristics of the household and education level of the mother ........................................................................................... 16 Figure 6: Education level of household head by consumption quintile ........................... 20 Figure 7: VUP Revenue, by source, RwF ...........................................................................38 Figure 8: VUP expenditure by activity, RwF ......................................................................39 Figure 9: Spending on social protection (excluding basic health and education), real 2010/11 RwF million ..........................................................................................................43 Figure 10: Government spending (actual) on social protection as a percentage of total government expenditure, 2004 – 2010/2011 ...................................................................44 Figure 11: Beneficiary numbers of ‘core’ social protection programmes, 2007 – 2009/10 ............................................................................................................................................47 Figure 12: PFM bottlenecks across the policy and budget cycle: impact on social protection ..........................................................................................................................50 iv Figure 13: Distribution of matched FARG beneficiaries by Ubudehe category and district ............................................................................................................................................88 Figure 14: Distribution of matched FARG beneficiaries by Ubudehe category in selected sectors in Kicukiro district ..................................................................................................89 Figure 15: Ease of consolidation of FARG Direct Support with VUP and Old Age Grant ..94 List of Boxes Box 1: A note on data sources for the systems assessment .............................................53 Box 2: Ubudehe targeting .................................................................................................57 Box 3: Methodological issues with the Baseline and Poverty Surveys ............................. 62 Box 4: Understanding fiscal decentralisation ...................................................................98 Box 5: Lessons learned from the Health Sector SWAp in Mozambique .........................114 v EXECUTIVE SUMMARY Context and objectives Rwanda’s social protection sector has grown rapidly in recent years, extending the coverage of new programmes but also developing core functions related to policy development and oversight. As a result, the sector has matured significantly; whereas only a few years ago it was characterised by a host of fragmented, un-coordinated and largely off-budget donor programmes, it is now very much government-owned (and funded) and increasingly coherent as an overall system of social protection. Rwanda is unique in the region, and indeed amongst developing countries, for the level of political commitment – backed by resources – but also the increasing sophistication of the policy dialogue and capacity of line ministries to effectively lead the sector. As one of the leaders in this area, however, it also faces a relatively new set of challenges for which there are no strong international examples from amongst developing countries. The overall objective of this Safety Net Assessment is therefore twofold. On one hand, it aims to bring together a disparate literature on the different programmes and assess each one in terms of design features, implementation features, and impacts. On the other, it aims to address some of these wider ‘second generation’ hurdles that will need to be overcome as the sector continues to grow and develop. Poverty and vulnerability Rwanda has experienced remarkable reductions in poverty over the last 10 years, with a notable acceleration in poverty reduction in the last 5 years in particular. Poverty has fallen from 59% in 2000/01 to 45% in 2010/11, while extreme poverty fell from 40% to 24% over the same period (NISR 2012). vi In spite of these achievements, the sheer magnitude of poverty facing the country over the last decade means that many Rwandans still live in extreme deprivation, often stuck in chronic poverty. Households who rely on farm wage labour continue to be amongst the very poorest, largely missing out on the overall trend in poverty reduction over the period. Many more are vulnerable to poverty in the face of shocks to production or consumption. The vast majority of the population rely primarily on rain-fed agriculture for subsistence, making them vulnerable to weather-related risks and the impacts of climate change. Although community-based health insurance covers the majority of the population, health risks are nevertheless still a concern for many. Social protection policy context Building on Rwanda’s Economic Development and Poverty Reduction Strategy (EDPRS 2008-2012), a National Social Protection Strategy (NSPS) was approved in January 2011. The Strategy sets out a comprehensive vision of the social protection sector over the short, medium, and long term. In so doing, it has provided a clear definition of the mission and objectives of the sector, as well as a delineation of which activities fall within the sector. The Implementation Plan that was adopted by the social protection working group in August 2011 provides the road map for putting the Strategy into practice. The objective of the sector is to: “build a social protection system that tackles poverty and inequality, enables the poor to move out of poverty, helps reduce vulnerability and protect people from shocks, helps improve health and education among all Rwandans, and contributes to economic growth”. In order to achieve this objective, the social protection sector is defined as having three separate dimensions. The first is the set of ‘core’ social protection activities, which comprise social assistance - the system of regular and predictable cash transfers that will provide income support to those living in poverty and vulnerable to falling into poverty - as well as the expansion of social insurance (mainly pensions and health insurance) and labour standards. The second is a set of broader social protection activities that are a means of ensuring access to other vii public services (particularly health and education) by enabling poor households to overcome the financial barriers that they may face. The final dimension is a set of social development initiatives and complementary activities, which are focused on providing additional support to vulnerable households and helping families graduate out of poverty. Expenditure and beneficiary numbers Total expenditure on social protection was over RwF40 billion in 2010/11, including social assistance, social insurance, social development initiatives and ‘complimentary’ services1. Adjusting for inflation, we see that spending across all areas of social protection has increased by more than 13 times in real terms between 2004 and 2010/11, over 14 times for the system of cash transfers alone, which compares well against an increase of almost 17 times for basic health and education. These real increases reflect not only an increase in the total resource envelope available to government; they also reflect an increased prioritisation in terms of the share of total government spending allocated towards social protection. Both the system of cash transfers and social insurance (mainly mutuelle subsidisation) each received 1.1% of the budget in 2010/11, while basic health and education received 10.6%. All the areas together represented over 15%. The number of beneficiary households from the main social protection programmes2 dipped initially from the level of 80,000 per year as the number of FARG Direct Support beneficiaries fell, however they have increased again as VUP continues to expand, reaching over 60,000 in 2010/11. In terms of the number of individuals reached by the core social protection programmes, in 2009/10 around 475,000 – or roughly 4% of the population – lived in households receiving assistance. 1 It does not include spending on other basic services such as health and education. 2 This excludes health insurance and focuses instead on programmes implemented by the sector: VUP Direct Support and Public Works, FARG Direct Support, and decentralized spending. viii Systems review: from design to implementation Vision 2020 Umurenge (VUP) VUP is the flagship social protection programme, which also plays a prominent role in the EDPRS. The overall goal of VUP is to contribute to the national target of reducing extreme poverty from 36.9% in 2005/6 to 24% in 2012, by accelerating the reduction of poverty in those areas where the programme is implemented. There are two components to VUP that are considered ‘core’ social protection: the first is direct support, which provides an unconditional cash transfer to eligible households, and public works which provides paid employment on productive community asset development projects. The third component is financial services, which provides micro- credit (and is not part of core social protection). The rationale is that by including these three different components, VUP will be able to appropriately address poverty of different groups of the poor: those who do not have labour capacity such as the elderly, disabled, or child-headed households who can benefit from direct support; while those who do have labour capacity and can benefit from public works employment, and those who are not in households eligible for social transfers through direct support or public works have access to micro-credit. As of 2011/2012 there are now a total of 120 sectors in the programme. The scale-up plan envisages that all sectors in the country will be covered for Direct Support by 2016, and by that time 240 sectors will have been reached for Public Works. There were 9,962 households receiving Direct Support in the fiscal year 2009/2010, and 61,355 work opportunities created under the public works component. VUP programme eligibility is based on the Ubudehe targeting approach, a community- based system where, through a participatory social mapping, the community ranks households according to where they think they fall on the poverty spectrum according to a set of global criteria. ix VUP has developed over time as the programme has expanded, and both the design and the implementation of the programme have been continually improved. Some of the outstanding issues are related to:  Delays in payment of transfers: related to delays in the targeting process as well as delays in disbursement and execution of funds.  Rationing of public works employment: due to budget constraints, not all eligible public works beneficiaries can benefit in any given year and the number of days of work is spread across households. As a result, only small cash benefits are provided to individual beneficiary households, thereby somewhat undermining the social protection objectives of the scheme.  Targeting: the Ubudehe targeting approach is now being used by a range of other sectors, which could end up undermining the effectiveness of the approach, as households recognise the great benefits that are attached with a particular Ubudehe status and angle to be included in a poorer category. This will need to be closely monitored. FARG Direct Support The Fond d’Assistance aux Rescapees du Genocide (FARG) was established by law in 1998, with the objective of collecting funds and distributing benefits to survivors of the genocide. According to the guidelines, Direct Support transfers are largely oriented towards orphans and elderly and disabled adults who lack other support systems. Beneficiary selection and identification is undertaken at the community level, facilitated by the Sector Executive Secretary. Although it is not poverty targeted explicitly, a list is drawn up of ‘needy’ genocide survivors who fit the criteria above. This list is then validated by the Sector Executive Secretary and by the district. x The number of beneficiaries increased from around 23,000 in 2001 to a high of 80,000 in 2007, although numbers have dropped significantly since then, falling back to 23,000 in 2010. This increase then decrease is likely to be at least partially driven by the demographic profile of beneficiaries indicated below; as the cohort of orphans going to school is now finishing it would be expected that numbers of eligible beneficiaries would decrease. Direct Support for disabled former combatants Disabled demobilised soldiers who are found to have particularly severe disabilities are provided with monthly direct support in the form of a subsistence allowance. Transfers are much larger (by up to ten times) than those under either VUP or FARG direct support, with the rationale that these former soldiers need additional support to ensure they can lead functional lives. The overall numbers of beneficiaries is, however, very low, with only around 2,500 beneficiaries in total. Social protection spending by districts Decentralised spending on social protection at district level – in the form of earmarked transfers - provides local authorities the ability to provide small grant opportunities for income generation. However, these are largely ad hoc and the values are extremely small. Social insurance  Public and formal sector workers have access to pensions, and health and accident insurance, through compulsory membership of Caisse Sociale and through health insurance schemes such as RAMA and MMI. However, although relatively wealthy self-employed people are also able to become members, coverage only extends to 4- 5% of the population. xi  Assistance with mutuelle (health insurance) membership is provided for roughly 10% of the population. As of the current fiscal year households will be selected on the basis of their Ubudehe status3. Social development activities  MINAGRI’s One Cow One Family has been considered to be socially protective, as it provides livestock which is then expected to provide additional income-generating potential for households who can sell milk or other dairy products.  Needy genocide survivors are catered for by FARG, which in addition to Direct Support also provides support housing, basic needs, and funding for income- generating projects (FARG’s additional programmes providing support for health and education are not considered to be part of social protection).  The Ubudehe programme provides support to one household in each village to enable them to take up an income-generating activity.  Refugees and returnees receive support financed by UNHCR and WFP, as well as by the Government of Rwanda through the National Refugee Council and other ad hoc transfers.  Finally, there is a wide range of other fragmented programmes providing grants for income generation, including programmes targeted at youth, women, etc. 3 However, it is important to note that the mutuelle system also provides an implicit subsidy to all members, since the insurance mechanism is not sustainable without contributions from overall government revenue. Furthermore, while in theory health insurance can be separated from other activities of the health sector, in practice it is just one part of the overall subsidisation of basic health service delivery. Rather, it should therefore really be considered as part of the wider approach to financing of the health service. xii Complementary services  OVCs are addressed through National Policy on Orphans and Vulnerable Children (2003), for which a strategy plan has been developed by MIGEPROF. A set of minimum standards has been established for the care and support of OVCs through the Minimum Package of Services for OVC. Gaps and overlaps/duplications Gaps in the current system are related mainly to the rate of scale-up and development of new programmes and to budgetary constraints. The main gaps are of two types:  Geographical, in the sense that VUP is now operating in 120 out of the 416 sectors out of the country.  Coverage in terms of poverty levels, since VUP only targets households in Ubudehe categories 1 and 2, reaching only a portion of the extremely poor households in the country. This is, however, due to budgetary constraints, and as a result the targeting process aims to identify those in the most need. While budgetary constraints will always remain in some form, the National Social Protection Strategy envisages a progressive realisation of the social protection floor over the next ten years. As such, the strategy and related implementation plan already envisage that the main gaps in coverage will be reduced over time. Given the clear strategic direction for the sector already existing, it is actually the current overlaps and duplications that require more urgent attention than the gaps. These overlaps are most obvious in the areas of direct support transfers (from VUP, FARG, and RDRC), housing support, and income-generating activities which should be the main responsibility of other sectors rather than a core focus of the social protection sector. In practice, however, these do not fall squarely inside of any one sector and xiii therefore tend to continue to be monitored mainly under social protection. Each of these programmes is implemented entirely separately, with separate uncoordinated funding streams, different target populations, and different mechanisms for selecting beneficiaries. Towards the next phase of the social protection sector: hurdles to overcome and a vision for the future Harmonisation and streamlining of programmes An analysis of FARG beneficiaries revealed that a simple integration of FARG Direct Support into VUP would exclude the vast majority (around 80%) of current FARG beneficiaries, many of whom are likely to be legitimately needy. Ultimately the targeting approach of VUP is not consistent enough with that of FARG, and the programmes currently serve different populations with slightly different objectives. In this case, ‘consolidation’ would actually mean discontinuing support under FARG Direct Support. Unless policy commitments are actively changed (for example, there is a clear policy decision that government is no longer obligated to support genocide survivors who are orphans, widows, or disabled), then it is not possible at the current time to simply fold FARG Direct Support into VUP. However, this is not to say that there are no other potential options. Given the demographic breakdown of FARG beneficiaries, with a portion of them being orphans whose eligibility for support will end by 2015 and the remainder being adults with disabilities and elderly, there may be scope to re-think the consolidation. Rather than forcing integration of two programmes with very different target groups and beneficiary selection processes in the short term, another option would be to consider consolidation in the medium term instead. The implementation plan has a feasibility study for an old-age grant as a major activity for this year, and this would be an ideal time to assess the feasibility of consolidating FARG and VUP Direct Support with an old xiv age/disability grant in future. The consolidation of these two programmes would be much more seamless than integrating FARG and VUP at present. In the short term, integration of FARG DS and VUP could still be partially achieved through the consolidation of delivery mechanisms. So, in other words, for the next couple financial years:  FARG DS and VUP targeting would still continue as they are now as separate processes.  In sectors where VUP is operating, FARG DS lists would be provided to VUP.  VUP would be responsible for the payment of beneficiaries, financial reporting, and monitoring. Decentralisation of social protection and institutional arrangements In many countries with well-established social protection sectors, including OECD countries that are often seen as the ‘model’ for decentralisation, social protection is actually a ‘shared function’ between central and local government because of concerns about both equity (across geographic areas) and accountability to the centre for delivery, especially of entitlements. Such concerns make it inadvisable to fully decentralize social protection down to district level. Rwanda’s social protection sector is currently a mix of an ‘agency’ deconcentrated model (FARG, VUP) alongside decentralisation to districts. For Rwanda, there is a clear overall desire for effective decentralisation, as articulated in the Decentralisation Strategy. However, for the social protection sector, some important questions going forward are: xv • Functions: what are the precise roles and responsibilities of MINALOC, agencies (FARG and RLDSF), and districts? Where should accountability lie for results at the end of the day – districts or RLDSF? • Financing flows: if funds flow directly to district, how will MINALOC and its agencies be able to ensure quality and equity in delivery? What levers will they have to intervene in under-performing areas? • Administrative control: Is the current assignment of functionaries consistent with the roles and responsibilities that have been laid out? In particular, does the responsibility for decisions over hiring/firing of district-level staff working on social protection (for example, VUP or FARG) fit with the responsibility for delivery? Inconsistencies between the functions, funds, and functionaries may arise if FARG/RLDSF is held accountable for results without having administrative control over personnel or flows of funds. Simple decentralisation may not be the most appropriate or effective approach for social protection; it may be that expanding the agency model might provide the best option for the sector, particularly as it expands into new programmes and more streamlined delivery mechanisms. Conclusions and recommendations from the Assessment What is clear from the assessment of the social protection sector in Rwanda is that very rapid progress has been made in the last 5 years, both in terms of the development and delivery of programmes, but also in the leadership, strategic direction, and coordination of the sector by MINALOC. These are achievements that are especially notable because they are not commonly found in other countries in the region, where donor-funded pilot programmes tend to be small in nature rather than national and government owned; scale-up happens slowly if at all; there is a strong lack of political commitment; implementation is hampered by ineffective targeting design; and activities are xvi dominated by uncoordinated and largely off-budget donor projects. By contrast with this norm, Rwanda’s social protection sector is a clear success story for the regio n as well as beyond. Appropriateness and relevance of social protection programmes There are clearly gaps in the level of current coverage, with the number of beneficiaries representing only about 4% of the population in total compared to extreme poverty levels of 24%. However, once VUP scales up as planned, that programme alone would be expected to cover around 18% of the population. In terms of the needs of particular groups identified in the poverty and vulnerability analysis, Rwanda’s social protection sector does generally cater for the needs of many of the core groups identified, at least in terms of intended coverage.  Extremely poor households with no labour capacity are included in VUP direct support. This should include the elderly who are no longer able to support themselves, female-headed households without adequate labour power, and households with persons with disabilities who similarly lack in adequate labour resources.  Extremely poor households with labour capacity (which would presumably include a large percentage of agricultural wage labour households identified as amongst the poorest and the most low-income agricultural households vulnerable to weather-related shocks) are eligible for VUP Public Works.  FARG Direct Support activities are designed to reach orphans, the elderly unable to work, and those survivors living with disabilities who are without other support. xvii  Other poor households are reached through social development programmes including VUP financial services, income-generating activities, and the provision of shelter. Whether these groups are actually reached in practice - and, importantly, whether other categories of the chronically poor and vulnerable are covered – is a matter for further evaluation of the sector. The Social Protection Strategy is explicitly based on a progressive realisation of a social protection floor, which envisages expanding coverage over time. The planned feasibility studies for a potential old age and disability grant should also ensure that these questions about appropriateness and coverage are addressed so that they can inform future directions for the sector. Issues related to the effectiveness of specific programmes The successes of VUP are numerous, not least the rapid development and implementation of the programme; the ability to put in place a set of operational procedures and staff training; the consistent and predictable scaling-up of the programme to a new lot of sectors each year; and, perhaps most importantly, the ongoing refinement of the process to increase efficiency and effectiveness. The Ubudehe targeting approach may not be perfect, but it seems to perform at least as well as other poverty-targeted approaches, and most importantly it has a very high degree of buy-in and ownership, so much so that it is implemented even beyond current VUP sectors so that other programmes such as the mutuelle de santé can utilise it. Districts also appear to have a high degree of ownership over social protection activities through VUP, mirroring the political commitment at all levels of government. These are achievements that are especially notable because they are not commonly found in other countries in the region, where donor-funded pilot programmes tend to be small in nature rather than national and government owned; scale-up happens slowly if at all; there is a strong lack of political commitment; implementation is hampered by ineffective targeting design; and activities are dominated by uncoordinated and largely xviii off-budget donor projects. By contrast with this norm, Rwanda’s social protection sector is a clear success story for the region as well as beyond. Issues related to programme design:  The definition of ‘graduation’ is problematic because it relies only on the subjective assessment of household’s status from one year to the next, and the same criteria for eligibility determine ‘graduation’. This latter issue is likely to undermine the ability of the programme to ensure long-term ‘graduation’: if households have their benefits removed as soon as they reach a higher standard of living they are likely to fall back down as soon as the support stops.  Ubudehe targeting may run into problems of effectiveness, especially now that Ubudehe categories are being used as eligibility criteria for more programmes.  The current levels of support provided through public works (in terms of the size of transfers received by households) are very small, and therefore the social protection provided by that component of VUP may in practice be smaller than originally envisaged in the programme design. Issues related to programme implementation:  The lack of a coordinated MIS system hinders the ability to effectively monitor implementation, although this is currently being addressed.  The timeliness and predictability of transfers under VUP needs to be addressed. Currently delays are caused by delays in implementing the Targeting Survey as well as lags in districts requesting disbursements. xix  Currently there is no way to account for unspent balances in VUP financial reporting, reflecting some deficiencies in both financial monitoring reports but also the budgeting process.  Practices in VUP public works of including non-eligible households in order to ensure quick implementation of projects needs to be addressed, as this undermines the targeting design of the programme.  As the initial public projects in terracing are exhausted, it will be a challenge for districts to identify projects that can be implemented to the requisite quality standard without significantly reducing the percentage of project expenditure going towards wages (as opposed to other inputs).  There is a need for a more systematic approach across programmes to fill the existing gaps in evaluation and ensure conceptual clarity across programmes when assessing performance and making inferences about appropriate policy responses. Higher-level strategic issues for the sector The discussion also raised a set of more strategic issues for the sector, many of which are inter-related. Harmonisation and consolidation of the sector:  The Strategy and Implementation Plan have set out the vision for ongoing harmonisation and consolidation of the sector, however this review found that before moving forward this will require further feasibility studies and dialogue on which groups will win and lose from this harmonisation. As the sector stands currently, ‘harmonising’ FARG Direct Support with VUP Direct Support would in practice mean that almost all current beneficiaries would simply be dropped, xx even if they are still legitimately needy according to FARG’s criteria. This is likely to contradict the current statutory commitments to genocide survivors as stipulated in the law on FARG.  Over the medium term, however, it is likely that there will be potential for a more seamless consolidation of programmes oriented around old age and disability benefits (which would encompass both current FARG and VUP Direct Support beneficiaries). Institutional arrangements and decentralisation:  Related to these issues of eventual consolidation of programmes is the question of the ideal institutional arrangements within Rwanda’s decentralised context. Finding the right balance of central and local engagement in social protection will require some further careful thought, however it is clear that the sector should not become fully devolved as is sometimes envisaged. Social protection is very likely to be best conceived as a shared function, and determining exactly which functions should be given to districts, as opposed to an implementing agency such as RLDSF, will require further study and discussion.  At the central level, it will be important for MINALOC to take an increasing role in the leadership of the sector. This, in turn, will require that staffing for the central policy unit is brought up to the levels indicated in the implementation plan, thereby allowing greater ministry inputs in terms of sector oversight and coordination through the Joint Sector Reviews and, crucially, in matters of public finance management, especially improving the consistency and coherence of the planning and budgeting in the sector. xxi Donor coordination: In many ways, the sector has already made some movement towards a SWAp, in terms of establishing joint donor and government reporting and review, and many of the large donors already operate at least partly through budget support. However, there are significant gains that can be recognized if this process continues. The Social Protection Strategy articulates the objective of moving towards a Sector Wide Approach (SWAp) which would: • Move more donor activity ‘on budget’ for greater harmonisation, coordination, and alignment. • Provide a mechanism for funding support going beyond VUP to include the sector as a whole. • Provide a coherent way to articulate total sectoral funding needs, financing gaps, and how donor support can meet these in a harmonised and coordinated way. Recommendations The flip side of the success and indeed Rwanda’s own ambitions for the sector is therefore precisely the identification of the next round of ‘second generation’ issues to be resolved. In order to address these issues related to specific programmes as well as the wider strategic issues for the sector, there are a set of recommendations as follows: (1) Ensure MINALOC has the proper human resource base for strengthened leadership and oversight of the sector First and foremost, in terms of process, the priorities are, firstly ensuring that MINALOC human resource capacities are strengthened. This is a pre-condition for the rest of the recommendations that follow: xxii (2) Improved oversight and coordination of the sector through the Joint Sector Review processes Reporting should be improved, so that there is analysis and interpretation of results in implementation, as well as concrete steps for addressing any issues involved. To cement these moves and to continue progressing further, it will be important to ensure that Annual Work Plans provide a realistic set of objectives to be achieved against which progress can be clearly measured. Technical working groups should be tasked with addressing the specific issues that have been identified across the design, implementation, and reporting of programmes: • Investigating targeting issues in VUP, including a quantitative and qualitative targeting study, and developing a ‘Graduation Policy’ for VUP, which could build on the findings of the targeting study. These issues of targeting and graduation need be examined together because they represent two sides of the same coin and yet in order to ensure effectiveness and sustainability graduation should not simply be defined on the same terms as initial eligibility.  Articulating a plan to ensure PFM issues in the smooth flow of funds from centre to districts for VUP are carefully monitored by the TWG and addressed as necessary. Other PFM issues to be addressed by the working group on Finance will be the way in which balances are carried forward and repayments on Financial Services component of VUP are budgeted and reported.  Developing an understanding on the expectations for the number of days of work (and hence total benefit values) to be provided to public works households , as well as the impact on beneficiaries if the relative costs of other inputs increases as activities other than terracing are undertaken. xxiii  Developing a coordination mechanism for IGA and shelter programmes under the ‘social development’ component of the sector.  Reviewing the evaluation framework across programmes. (3) Establish a process for the development of deepened joint donor/government coordination Mechanisms of dialogue should be well articulated, to ensure that the SWAp that emerges has built-in mechanisms for dealing with any bumps in the road that may arise. There is a need to be clear from outset about expectations of donors for participation in SWAp, and setting clear milestones for their harmonisation of aid instruments. (4) Lead a process of strategic thinking for the medium- and long-term development of the sector, in line with the vision in the Strategy and Implementation plan. This will include:  In the short term, establishing the potential gains from and feasibility of delivering FARG Direct Support (using the current targeting processes) through RLDSF.  Undertaking the feasibility assessment for the old age grant as envisaged in the work plan for the year, including the micro-simulation of different programme options (looking at impacts and financing requirements) as well as a functional review. This will serve as the basis for:  Dialogue on government commitments (statutory or otherwise) to different vulnerable groups. The key question to examine is: What is the best way to balance the desire for consolidation of programmes with the desire to continue to reach groups currently covered? xxiv  The clarification of strategic objectives with respect to decentralisation of social protection sector, leading to the development clearly articulated and consistent function assignments, financial flows and administrative arrangements. (5) Translating this movement towards the implementation of the strategic vision and the increasing coordination and coherence of the sector into improvements in the planning and budget/MTEF processes This will require engagement with MINECOFIN on how best to ensure coordination and coherence within their existing budget process. It is likely that there should be both a short-term approach to simply increasing coordination by MINALOC of its relevant agencies (especially FARG and VUP), as well as a medium-term view of how the budget and MTEF classifications – and processes of accountability – can be better aligned with the definition of the sector as it evolves over the next several years. xxv 1. Introduction 1.1 Background and context 1. Rwanda’s social protection sector has been steadily developing over the last several years. It featured as a major part of the EDPRS, with the Vision 2020 Umurenge Programme (VUP) one of the three major flagship programmes of government. Public expenditure on the sector has been increasing, reaching 2.8% of total expenditure in the 2009/10 budget4. 2. As in many countries, one of the major challenges has been in defining what exactly social protection entails, since different stakeholders have different views about how broadly or narrowly the sector should be defined. Despite the good progress achieved in the sector, further challenges remain related to the fragmentation of social protection activities, some with their own targeting process, benefit package, and implementation procedures. The Ministry of Local Government (MINALOC) is charged with overall responsibility for the sector and accountability for results, though a few other programs remain somewhat dispersed across other ministries namely: the Ministry of Disaster Management and Refugees Affairs, Ministry of Gender and Family Promotion, Ministry of Agriculture and Ministry of Health and agencies (including FARG, Rwanda Local Development Support Fund – formerly CDF, VUP and Ubude programs) and coordinating bodies (National Council for Refugees, National Council on Disability, Demobilization and Reintegration Commission). The agencies and coordinating bodies are affiliated to MINALOC as their parent ministry. Donor activities are reasonably well coordinated through the Sector Working Group (SWG) and donor coordination meeting, but these nevertheless add a further layer of complexity in coordinating efforts. 3. Given these challenges around delimiting the social protection sector in a coherent and coordinated manner, a major milestone in the development of the sector 4 Government of Rwanda (2010) Social Protection Public Expenditure Review. Kigali: MINECOFIN. 1 is therefore the elaboration of the National Social Protection Strategy (NSPS) and the associated Implementation Plan (for 2011-2016). The strategy sets out a clear definition, separating those activities that fall within the boundary of the sector and those which are complementary – but separate – activities. 4. The overall objective of the strategy is to build a social protection system that tackles poverty and inequality, enables the poor to move out of poverty, helps reduce vulnerability and protect people from shocks, helps improve health and education among all Rwandans, and contributes to economic growth. It therefore takes a holistic and integrated approach, consisting of four main areas, including (i) a comprehensive system of cash transfers based on existing CTs in VUP, FARG and the ex-combatants program; (ii) the extension of contributory social security and labour standards; (iii) scaling-up cross-cutting complementary services (including gender mainstreaming, savings and cooperative services, community health, etc.); and (iv) the integration of interventions that allow timely responsiveness to risks and shocks. 5. The strategy and implementation plan also set out a vision for increasing harmonisation of activities, in particular integrating several of the FARG programmes and the transfers to ex-combatants into VUP. The aim is to reduce the duplication and overlap between these programmes and to streamline the targeting, beneficiary selection and identification, and payments process, thereby eliminating unnecessary administrative costs wherever possible. 1.2 Objectives of the Assessment 6. With the strategy and implementation plan now in place, outlining a concrete vision for the development of the sector over the next few years, the Concept Note for the Social Safety Nets Assessment5 recognised that the most pressing issue at the moment is to further work out the details of how these objectives will be achieved. Despite the notable achievements in the sector in recent years, there are a number of 5 See Annex A for further details. 2 weaknesses that will need to be overcome in order to put the strategy and implementation plan into action. With this context in mind, the objectives of the SSN Assessment as developed in the Concept Note are to:  Review the landscape of existing SSN programs, interventions and activities . This will build on existing work in the sector and will focus on the details of selection criteria, benefit packages, delivery mechanisms, costs, and funding sources.  Assess institutional arrangements and related programming for existing SSNs at the national and local government levels as well as local and international NGOs. Of particular importance are harmonisation arrangements, planning and budgeting processes, financing modalities, implementation arrangements, and monitoring and impact evaluation.  Assess central-local government fiscal relations with regard to financing implementation of SSNs. This is particularly important given the ongoing changes in decentralisation which are underway.  Ascertain synergies, complementarities and overlaps in existing SSNs at the policy level as well as implementation and delivery mechanisms.  Review impact evaluation arrangements including monitoring and reporting frameworks and the MIS systems in place.  Make a set of recommendations to help the sector overcome existing weaknesses and shortcomings from the level of policy through to service delivery. 3 7. Previous analytical work in the sector including the Public Expenditure Review (PER) and analysis feeding into the strategy and implementation plan provide a solid starting point in many of these areas. The SSN Assessment will therefore build on these previous efforts and will focus on those areas that have not yet been explored in detail including those that are deemed critical to putting the NSPS and its implementation plan into action. 8. Commensurate to the objectives outlined in the concept note, some of the issues that emerged as top priorities for many stakeholders were specifically related to the need to:  Carefully think through the implications of potential integration of some FARG programmes and transfers to demobilised disabled combatants under the VUP implementation structures, specifically in terms of managing beneficiary selection and identification and payment mechanisms, as well as financial management processes.  Understand how plans for further fiscal decentralisation will impact social protection spending and, conversely, how the sector might best engage with the MINECOFIN in ongoing policy discussions on this area.  Consider the overall context of accountability within the decentralised system (for example the nature of performance contracts at district level - Imihigo), and what this means for incentives and accountability for service delivery on the ground, particularly, given a recent resolution from the Imihigo 2010/11 review to give more focus to interventions that respond to the citizenry welfare, key of these being social protection interventions.  Get a sense of the extent to which social protection is understood as a concept at district levels, and the extent to which district officials perceive the sector as a coherent whole. 4  Address the current gaps in impact evaluation in the main social protection programmes.  Assess the steps that will be needed for a joint donor funding arrangement to be put in place (as per the implementation plan), including progress that will need to be made by MINALOC in building capacity as well as what donors will need to do. 1.3 Note on methodology Scope 9. Although the focus is on social safety nets, the scope of the Assessment is actually consistent with the wider definition of the sector as elaborated in the Social Protection Strategy. This defines the social protection sector as the system of predictable cash transfers that will provide income support to those living in poverty and vulnerable to falling into poverty, and in addition to this it also includes social insurance, social development initiatives, and complementary programmes6. Programmes identified as part of the sector include: VUP direct support grant, VUP public works, FARG, RDRC, and the Child and Refugees Support Programme. The Assessment will generally not cover programmes implemented by other sectors, such as grants and scholarships in education, health insurance, or youth programmes. 6 While it excludes the role of social protection as a cross-cutting aspect enabling better access to other public services. 5 Defining safety nets As with many aspects of social protection, the definition and nomenclature has at times been somewhat contentious, with different institutions having different understandings of what policy areas and programmes are included and how activities should be labelled. According to standard World Bank definitions of Social Safety Nets, it includes “non -contributory transfers targeted in some way to the poor or vulnerable”. This includes cash transfers (whethe r conditional or unconditional, non-contributory pensions, public works, and family/child allowances) that are sometimes referred to as social assistance by other stakeholders. This definition of safety nets also includes wage/employment subsidies, school vouchers/scholarships, and fee waivers. For the purposes of the current assessment, however, the safety nets definition is not adhered to rigidly. This is because in order for the assessment to be most relevant to the Rwandan context, it should follow the definition and classifications that are used by the Social Protection Sector Working Group, as articulated in the Social Protection Strategy. 10. It does, however, include MIGEPROF’s Orphans and Vulnerable Children (OVC) activities and MINAGRI’s One Cow One Family Programme since these areas in which the Sector Working Group has not so far been able to obtain much information and the SSN Assessment could help shed some light. Other programmes that will also be included in the Assessment (although not in as much detail as the core social protection programmes) are the VUP Financial Services component, since the implementation and budget process is tied in with other VUP activities; FARG income generation since this may be integrated into VUP activities, and the Ubudehe community support programme, since this is likely to be included in any joint funding modality of donors. Data collection 11. The data collection phase involved the following steps:  Scoping exercise at district level (June 2011): involving initial visits to a couple districts to get a better sense of who is involved, processes related to budgeting and planning, beneficiary targeting and identification, budget implementation, and reporting. The objective of the scoping exercise was to gain an initial understanding of the situation at district level, in order to ensure that the data collection instruments used is well designed. 6  Collection of central-level data (July – August 2011): There are still some major gaps in the data available at central level, particularly in terms of FARG, RDRC, and refugees and returnees. Data collection involved collection of – where available – disaggregated budget documentation by programme, details on numbers of beneficiaries and benefits paid, monitoring reports, and any guidelines available on targeting, implementation, or financial management.  Initial round of analysis and development of detailed interview guides (July 2011): based on the first two activities, an initial round of data analysis was undertaken, so that the interviews in the main data collection mission were as effective and focused as possible.  Main data collection mission: (July 2011) This included interviews with central- level stakeholders at MINALOC, MINECOFIN, VUP, RDRC, FARG, Social Security Fund, as well as development partners. It also included a fieldwork component, including a sample of districts and sectors.  Follow-up data collection (August – October 2011): To fill in any remaining gaps and validate the initial findings.  Final presentation and validation (November 2011): A stakeholders meeting held jointly by GoR and the Bank team to get final inputs before report finalization and publication.  Internal peer review within the World Bank: The report received input from selected lead experts on social protection in the World Bank to benefit from international lessons emerging from similar and related programs as those outlined in the NSPS. The research process was therefore fairly iterative, to ensure enough time and space for the researchers to tease out the important underlying issues. 7 1.4 Structure of the report 12. In order to provide the backdrop for the subsequent analysis, Chapter 2 will provide an overview of the sector, including the policy and strategic framework; key actors and their roles and responsibilities; and a review of previous analysis already undertaken. Chapters 3 and 4 then provide a detailed review of the current social protection system, with Chapter 3 providing information on each of the programmes including objectives, targeting approaches, benefit packages, management and implementation arrangements, and monitoring and evaluation approaches, while Chapter 4 presents summary information in terms of financial flows and brief assessment of current strengths, gaps in coverage, and areas of duplication and overlap. Throughout Chapter 3 ‘red flags’ will be identified in the text; these are issues that will need to be addressed in the current programmes to ensure they are as effective and efficient as possible. Chapter 5 then provides a discussion of four of the most important hurdles facing the sector as it moves into its next phase, including the harmonisation/consolidation of programmes; establishing the optimal level of decentralisation for the sector; institutional arrangements; and establishing a formal Sector Wide Approach and common funding mechanism. Finally, Chapter 6 provides conclusions and recommendations for the way forward. 8 2. Poverty, risk, and vulnerability in Rwanda 13. Rwanda is a small land-locked country, bordering Uganda, Tanzania, the Democratic Republic of Congo, and Burundi. It has a population of around 10.7 million (estimated7), but given its small size (26,338 square km, around the size of the state of Maryland in the United States) it has the highest population density in Sub-Saharan Africa, at 430.64 people per square kilometre. The population remains largely rural, with around one million people residing in the capital city Kigali, although urbanisation has been increasing in recent years. This population density is all the more important, given the mountainous terrain that characterises most of the country, which leads to many small plots terraced on mountainsides that are vulnerable to landslides. 14. The country has made enormous strides since the devastation of the genocide in 1994. It adopted a new Constitution in 2003, and has implemented two successive Poverty Reduction Strategy Papers, the first in 2001, and the second, the Economic Development and Poverty Reduction Strategy in 2007, both of which have emphasised macro-economic and structural reforms to lay the groundwork for economic growth, poverty reduction, human capital investment, and good governance. An important reform included efforts in decentralisation, which involved administrative re- organisation. There are thirty districts with elected mayors and councils, which are organised into four provinces (Northern, Southern, Eastern, and Western) and the city of Kigali. Each district is further divided into sectors. 15. Economic growth has been encouraging at 7.5% in 2010, performing above the East African Community and Sub-Saharan African average. 7 Based on population projections; a new census round is currently underway. 9 nsification Program (CIP) and the Saharan Africa (SSA) (Figure 2). During 2010 sbandry, Water Harvesting and the services and industrial sectors rigation (LWH) Project, the latter progressed in their growth recovery, while supported with donor funds. growth in the agricultural sector slowed Figure through egic focus on agriculture, down marginally. 1: Real GDP growth per annum in Rwanda compared to EAC and SSA Average public investments, has Figure 2. Rwanda Real Growth Compared to EAC and SSA ed to marked productivity (Percent) 12.0 and solid agriculture growth rates nt years. However, in order to 8.0 ese productivity increases in the nd in order to fully realize the tential for the agriculture sector, a 4.0 of challenges would need to be . These challenges include the 0.0 stronger focus on: (i) Reducing cy on rain-fed agriculture through Source: World Development Indicator (WDI) and e of different models of irrigation; International Monetary Fund erosion control and integrated soil management; (iii) Diversifying was remarkably stable, given the difficult 16. Rwanda has experienced remarkable reductions in poverty over the last 10 years, e production, in particular external post-crisis environment and al export goods, for a notablein withexample acceleration in poverty reduction in the last 5 years in particular. It is not - horticulture and only consumption flowers, dependent land-locked country. This wasin health, education, (iv) poverty that has fallen; significant achievements the skills profile of people and other of poverty dimensions mainly have also achieved been recorded, through a prudentputting fiscalRwanda on track for in agriculture, to foster the stanceThis with meeting many of the MDGs. is an strong especiallyfocus priority on feat impressive considering the of increased agricultural off-farm expenditures, assisted by continued high devastation wrought ent such as agro-processing and by the 1994 genocide. grant financing from donors. Government lue chain activities, and (v) followed a large public investment program g a market-based food crop on a number of the strategic key sheer projects, 17. In spite of these achievements, magnitude ofall in poverty facing the n system to contribute to country- the infrastructure sector. But to date this is d security. Continued country over the last decade means that many Rwandans still live in extreme agriculture largely financed with domestic revenues and outlined above rough the channelsdeprivation, often stuck in chronic debt remains poverty. Many more at manageable areInflation vulnerable to poverty in the levels. it agriculture growth and increase face of shocks to production or consumption. The vast majority of the population rely vii | P a g e primarily on rain-fed agriculture for subsistence, making them vulnerable to weather- related risks and the impacts of climate change. Although community-based health insurance covers the majority of the population, health risks are nevertheless still a concern for many. 18. Given that the nature of poverty, the underlying causes of deprivation, the susceptibility to adverse shocks, and the types of positive (or negative) coping mechanisms households have available are of central importance to the design of appropriate social protection policy responses, this chapter provides an overview of the 10 poverty context in the country. Unfortunately a panel survey has not been undertaken (which would allow an analysis of chronic versus transient poverty), but the country is nevertheless rich in a range of other data sources which can be used. The following section will first look at the broad trends in poverty and extreme poverty over time and space. This is followed by a discussion of the correlates of poverty to better understand the characteristics of poor households and finally a discussion of risks, vulnerability, and coping mechanisms facing households. 2.1 Trends in poverty over time 19. As illustrated in the figures below, poverty has fallen from 59% in 2000/01 to 45% in 2010/11, while extreme poverty fell from 40% to 24% over the same period (NISR 2012). There is still some regional variation in poverty across provinces, but it is interesting to see that the Northern Province, which was initially one of the poorest, has now reduced poverty to the national average. 20. The depth and severity of poverty have also fallen over the same period, although the depth of extreme poverty in 2010/11 was still 26%, meaning that on average the extreme poor have consumption that is 26% below the extreme poverty line. Considering that the extreme poverty line is based only on the bare minimum of caloric needs for survival (so that households must forgo some of this food requirement in order to meet other essential non-food needs), it is clear that many of Rwanda’s poor continue to live at levels of extreme deprivation. In addition, the distribution of consumption is relatively flat, meaning that there are many households living just above the poverty line who are also vulnerable to poverty. 11 Figure 2: Poverty and extreme poverty incidence, 2000/01 to 2009/10 Source: NISR (2012) using data from EICV I, II, and III. 21. Within these broad provincial groupings, however, it is important to note that there is considerable variation across districts. The greatest concentration of poverty is in the south and southwest including districts in Southern and Western Provinces. Figure 3: Incidence of poverty 2010/11 by district Source: NISR (2012) from EICV III data. 12 22. While inequality increased over the 2000/01 to 2005/06 period, it actually decreased between 2005/06 and 2010/11, with the poor increasing their consumption more quickly than the rich. This is illustrated in the growth incidence curves for the two periods below, which shows the annualised rate of growth in consumption for each consumption percentile. Figure 4: Growth incidence curves, 2001/2 – 2005/6 and 2005/6 – 2010/11 Other dimensions of poverty 23. Consumption poverty is also associated with poverty in other dimensions, as shown in the tables below. For example, net primary enrolment increased nationally from 86.6% to 91.7% between 2005/6 and 2010/11 but children in the bottom consumption quintile have enrolment rates that are nearly ten percentage points lower than those in the richest. Nearly all children now attend at least some primary school, so the enrolment rates below largely reflect the fact that some children start late and drop out early. The basic cycle of primary education is free, so these disparities across socio- economic groups suggest that there are still important financial barriers to access to allow children to complete primary. 13 Table 1: Education indicators, EICV2 (2005/6) and EICV3 (2010/11) School attendance Literacy rate (15 to Net Primary Net School Tertiary education (6 years +) 24 years) enrolment enrolment (16 to 35 years) EICV2 EICV3 EICV2 EICV3 EICV2 EICV3 EICV2 EICV3 EICV2 EICV3 Rwanda 78.7 83.2 76.9 83.7 86.6 91.7 10.4 20.9 1.2 2.6 Q1 71.3 77.7 66.3 75.6 79.9 86.9 2.2 8.6 0.1 0.1 Q2 75.0 80.2 72.9 80.7 86.3 91.4 5.6 13.0 0.0 0.3 Q3 77.9 82.6 77.2 83.6 88.1 93.0 9.3 18.7 0.1 0.5 Q4 80.6 83.7 80.3 86.0 89.8 93.7 14.2 24.3 0.3 0.7 Q5 87.2 90.6 84.2 88.9 91.2 95.7 21.3 39.8 3.8 8.2 Source: NISR (presentation) from EICV II and III. 24. Similarly, there have been remarkable improvements in health indicators over the last decade, with under-five mortality falling by half in just five years, from 152 in 2005 to 76 in 2010, and the maternal mortality rate plummeting from 1071 in 2000 to 487 in 2010. 14 Table 2: Childhood mortality rates Figure 8.1 Trend in Childhood Mortality Rates Deaths per 1,000 live births 180 160 152 140 120 103 100 86 80 72 76 62 60 49 50 43 40 37 34 28 27 23 27 20 0 Neonatal Postneonatal Infant Child Under-5 mortality mortality mortality mortality mortality 2005 RDHS 2007-08 RIDHS 2010 RDHS Source: RDHS 2010. 8.3 SOCIOECONOMIC DIFFERENTIALS IN CHILDHOOD MORTALITY Results presented in Table 8.2 and Figure 8.2 show that childhood mortality in Rwanda varies significantly by the socioeconomic characteristics of households and mothers.1 Mortality in urban areas is generally lower than in rural areas. For example, infant mortality in urban areas is 55 deaths per 1,000 live births compared with 62 deaths per 25. 1,000 liveHowever, there births in rural are areas. The still differences urban-rural across gap is wider for wealth neonatal quintiles, mortality (21as shown deaths versusin the 31 deaths per 1,000). Differentials in mortality by province are also substantial, particularly in the under-5 mortality rates. The City of figure below. Children whose mothers have had no schooling are almost twice as likely Kigali has the lowest rates of neonatal mortality (21 deaths per 1,000 live births) and under-5 mortality (79 deaths per 1,000 live births). The highest neonatal mortality and infant mortality rates are found in the North province (39 to die before the age of five than those with secondary education or higher, and deaths and 71 deaths per 1,000 live births respectively), whereas the East province has the highest rate of under-5 mortality (125 similarly the deaths per 1,000 mortality live births). rate for those in the lowest wealth quintile is 119 compared to 75 for those in the highest. 1 To have a sufficient number of cases to ensure statistically reliable mortality estimates, rates presented in Tables 8.2 and 8.3 are calculated for a 10-year period. Infant and Child Mortality • 103 15 Second 32 25 57 49 103 Middle 25 35 61 46 104 Fourth 31 35 66 41 104 Highest 24 26 50 27 75 Computed as the difference between the infant and neonatal mortality rates Figure 5: Under-five mortality by socio-economic characteristics of the household and Figure 8.2 Under-5 Mortality Rates by education level of the mother Socioeconomic Characteristics RESIDENCE Urban 81 Rural 105 EDUCATION No schooling 125 Primary 99 Secondary and higher 63 WEALTH QUINTILE Lowest 119 Second 103 Middle 104 Fourth 104 Highest 75 0 20 40 60 80 100 120 140 Deaths per 1,000 live births Source: RDHS 2010. RDHS 2010 2.2expected, As mortality declines Understanding markedly poverty: who asare mother’s education increases. Children born to mothers with the poor? no schooling have the highest mortality rates. According to the survey results in Table 8.2, the infant mortality rate 26. Beyond these broad trends, it is important to try and tease out the characteristics of the poor and vulnerable, to better understand who is poor and what 104 • Infant and Child Mortality may contribute to either being stuck in a poverty trap, or potentially moving out of or falling into poverty. Vulnerable groups 27. The Social Protection Policy of (GoR 2005) identifies categories of vulnerable groups to be: genocide survivors; orphans; Other Vulnerable Children (OVCs); widows; prisoner’s families; youth (16-25s); demobilised soldiers; repatriates and refugees; elderly; victims of catastrophes; destitute people (those who cannot meet their primary needs such as accommodation, food clothing, and basic health care); and historically marginalised groups (especially the Batwa). 16  Child-headed households: In 2005/6 only 0.2% of the child population lived in child-headed households (around 10,000 children), and this percentage had fallen by half since the 2000/1 survey round. Given the very small sample size, it was therefore not possible to assess their poverty levels in a robust manner but this group – though small - is expected to continue to be especially vulnerable.  Orphans: In 2010, 13% of children were orphans; 2% were double orphans, 9% were paternal orphans, and 3% were maternal orphans. According to the 2005/6 EICV, double orphans, paternal orphans not living with their mother, and maternal orphans were actually less likely to be poor, seemingly as a result of frequently being adopted or fostered by better-off relatives8. However, orphans were found to be less likely to attend school than non-orphans, emphasising that living in a non-poor household does not necessarily ensure that children’s other rights to health and education will be met9.  Female- and widow-headed households: In 2010/11 around 28% of the population lived in female-headed households. These were more likely to be poor than male-headed households, but only slightly (41.5% in poverty compared to 39.7%).  People living with a disability (PWDs): The 2002 census estimated that there were roughly 300,000 disabled persons in the population10. There is no data on their poverty incidence, however economic participation rates were found to be higher for the disabled than the general population11. 8 Note that in 2005/6 orphans were a larger percentage of the child population – 18% instead of 13% currently – as a result of the fact that the cohort of children orphaned by the genocide was younger at that time. 9 In terms of OVCs, it is important to recognise that a more nuanced understanding of poverty is required, since they are dependent on their caregivers and service providers for needs on many dimensions and are vulnerable even within households that may not be poor on income dimensions. 10 A more comprehensive survey of people living with disabilities is currently scheduled, which should add significantly to the evidence base on the living conditions and needs of this group. 11 This cannot be explained without further data, however it may be destitution could play a role in forcing a higher degree of labour participation in order to ensure survival. 17 28. Finally, it is important to also note the particular case of genocide survivors, a vulnerable group that is unique to Rwanda as a result of the events of 1994. The 2007 census of survivors12 estimates that there are around 310,000 individuals, of which 58% are women. Given the need for extreme sensitivity in the collection and interpretation of data related to this group, it is not possible at this time to make a full comparison in terms of poverty levels of genocide survivors as compared to the rest of the population. It is, however, clear that survivor households are more likely to contain orphans (21% of the total survivor population are orphans, and only 28% of survivor children have both parents alive), and disabled (7.3% - estimated to be around double that found in the 2002 census nationally). Survivors are also disproportionately young, with 66% of them aged 13-35. Livelihood categories 29. Beyond these categories of vulnerable groups, poverty is very closely associated with a household’s livelihood category. One of the most consistent findings from both quantitative and qualitative sources over the last ten years has been the high rates of poverty and extreme poverty amongst agricultural wage labour households, as reflected in the table below for the last two survey rounds. For example, for those households that receive most of their income from farm wage labour or who are diversified in income sources but still rely on farm wage work for 30% of their income or more, poverty stands at 76-77% compared to 45% nationally. 12 NISR (2007) Recensement des Rescapé. 18 Table 3: Poverty incidence by main source of income Source: NISR (2012: 19). 30. Similarly, a recent Comprehensive Food Security and Vulnerability Assessment (CSFVA) from 200913 found that three livelihood categories were consistently the most poor and vulnerable across a range of indicators including income, consumption, and the frequency and diversity of food consumption. These three categories were low- income agriculturalists (who have very small plot sizes) agro-labourers (who had some own small production and also worked as daily wage labour), and marginal livelihoods (a collection of household depending on assistance, remittances, and other unspecified activities). 31. Nationally, 4% of households were found to have a poor Food Consumption Score (FCS), 17% have a borderline FCS and 73% have an acceptable FCS. These same three livelihood categories make up 73% of all those with a poor FCS, although they make up only 46% of the total population. 32. As in the EIVC data in the table above, the CSFVA also found that households with only one activity are far more frequent amongst households with poor FCS (45%) compared to those with borderline FCS (37%) or acceptable (31%). The lack of diversity 13 NISR (2009) Rwanda Comprehensive Food Security and Vulnerability Analysis and Nutrition Survey. 19 in livelihood options – outside of agricultural wage labour - therefore emerges as an important theme in poverty and vulnerability. Correlates of Poverty 33. We can also estimate the ‘determinants’ of poverty (or, more accurately, the correlates of poverty) in a multi-dimensional model, using EICV3 data from 2010/11. Before deciding on which variables should be included in the model, it is important to understand which variables are most associated with poverty. Looking at the distribution of different variables across consumption quintiles helps to provide a good picture of the characteristics of the poor versus those who are better off. The following variables were identified as good candidates for inclusion in the model:  Highest level of education of the household head: The figure below illustrates how those in the poorest quintile are far more likely to have had only some primary or less (82% compared to 46% in the richest), while those in the richest are over sixteen times more likely to have had vocational, secondary, or higher education (2% of the poorest quintile compared to 33% of the richest). Figure 6: Education level of household head by consumption quintile 20  Type of Lighting source: The table below shows that electricity is almost exclusively a luxury for those in the richest quintile, while oil lamps decrease in frequency in as households become better off and firewood is more prevalent amongst poorer quintiles. Table 4: Lighting source by consumption quintile, national Quintile Electricity Oil lamp Firewood Lantern Piles Other Lowest 0.7 9.05 40.85 15.03 21.4 14.31 Second 1.36 12.25 20.08 22 23.04 14.52 Third 3.95 18.12 17.83 24.97 21.97 16 Fourth 11.18 27.38 13.88 24.44 21.66 26.92 Highest 82.81 33.2 7.35 13.55 11.92 28.7 Total 100 100 100 100 100 100  Type of toilet: Flush toilets are mainly found amongst the richest quintile of households, while households with no toilet facilities are most common amongst the poorest.  Household durable assets: Radios follow a somewhat surprising pattern, with 22% of those in the second quintile owning them, but few in the poorest, third, or fourth quintile. Bed ownership rises steadily with consumption quintile, as does bicycle ownership. Table 5: Household durable ownership distribution across quintiles Quintile Radio Bed Bicycle Lowest 2.46 11.02 5.37 Second 22.02 16.28 14.02 Third 5.75 20.07 21.15 Fourth 11.25 23.85 30.07 Highest 58.53 28.78 29.39 Total 100 100 100  Livestock: Livestock ownership increases, as would be expected, across the first four quintiles, but then decreases slightly in the richest. Those in the richest quintile are more likely to be in skilled non-farm work, and therefore less reliant on agricultural production than those in other consumption quintiles. 21 Table 6: Distribution of livestock ownership across consumption quintiles Quintile Cattle Sheep Goats Chickens Lowest 11.24 13.59 14.41 14.41 Second 17.13 19.35 20.16 20.16 Third 21.65 22.36 22.5 22.5 Fourth 26.73 25.12 24.82 24.82 Highest 23.26 19.58 18.11 18.11 Total 100 100 100 100 Pulling these variables together into a regression, the table below shows the results from the model, which uses the log of per-adult equivalent consumption as the dependent variable. Note that for the categorical variables, the score for the first category is zero since this is the reference category. The other categories can be interpreted in reference to this, so for example positive coefficients mean that other categories are better off compared to the reference category, while negative coefficients mean that households in the other categories are worse off than the reference category. The signs of the coefficients are all as would be expected: higher levels of education are associated with higher consumption, while household size is associated with lower. Households using electricity for lighting are better off than all other types of lighting. Ownership of assets (livestock, radio, bed, bicycle) is all associated with higher consumption. Having a household member work in agricultural wage work is associated with lower consumption (as would be expected), while wage work in the non-farm sector and participation in some kind of household non-farm enterprise (of any size, including micro-enterprise) is associated with higher consumption. 34. The variables all have the expected sign and are generally highly significant. The explanatory power of the model overall (adjusted R-squared) is .52, which is fairly consistent with other models of this nature employed elsewhere. 22 Table 7: Correlates of poverty – regression results Dependent variable: log of per-adult equivalent 14 consumption Coefficient Standard Error Education level of household head No education (reference) 0 Some primary .033** .011 Completed Primary .053*** .014 Vocational Education .182*** .026 Secondary Education .388*** .025 Higher education 1.061*** .043 Household size -.120*** .002 Lighting Source Electricity (reference) 0 Oil lamp -.662*** .025 Firewood -1.091*** .028 Lantern -.868*** .023 Piles -.948*** .023 Other -.785*** .027 Type of Toilet Flush (reference) 0 Pit latrine with solid slab -.759*** .046 Pit latrine without solid slab -.750*** .047 Other -.644** .225 No toilet -.787*** .050 Has radio .325*** .096 Has bed .178*** .011 Has bicycle .175*** .015 Has cattle .114*** .011 Household member works for Ag wage -.169*** .010 Household member has non-farm enterprise .105*** .010 Log of land cultivated (acres) .058*** .004 Province Kigali city (reference) 0 0 Southern -.288*** .023 Northern -.190*** .022 Western -.137*** .023 Eastern -.202*** .023 Constant 13.30*** .052 N = 13,949. 14 * p<0.05, ** p<0.01, ***p<0.001 23 2.3 Risk and vulnerability 35. A vulnerability assessment15 conducted in 2005 outlined the main sources of risk for Rwandan households. Since roughly 90% of the population relied on agriculture for their livelihood, agricultural risks were found to be the most significant, although other risks including those to health are also important. Agricultural and agro-climatic risks 36. Poor rainfall and land slippage were the most common shocks that occurred, and the 2009 CFSVA found that drought, irregular rains, or dry spells caused a reduction in consumption in 23% of households, although this varied very widely by district16. Health risks 37. When an illness occurs, households are faced with a double shock: they need to pay for immediate medical expenses while at the same time coping with the loss of income in the case of adult members of the household. Health risks are therefore also presumed to be of substantial importance, although more concrete evidence on the extent and impact of health risks is required. According to the 2009 CFSVA, 9% of households reported facing a shock from the illness/accident of a household member. 38. Access to health care has increased remarkably in recent years as a result of the expansion of community-based health insurance (the mutuelle), with almost 70% of the population covered in 2010/1117. However, this masks wide variation in coverage across 15 Prywes and Verwimp (2005) Rwanda: Assessment of Risk and Vulnerability and of Vulnerable Groups. 16 This variation is also consistent with the findings of the livelihood zoning activity undertaken by USAID(2011). 17 Note that the RDHS of 2010 reports a slightly higher percentage, with 77.8% of households having at least one member covered by insurance. This could be due to methodological differences, since the two surveys use slightly different indicators (individuals covered vs. percent of households with at least one member covered). The RDHS however confirm the general trend that households in the poorest wealth quintiles have lower coverage. 24 consumption quintiles, with only half of those in the poorest quintile having insurance compared to nearly 85% in the richest. Table 8: Health insurance coverage and physical access to health centres, EICV2 (2005/6) and EICV3 (2010/11) Times (min) to reach health Health insurance coverage (%) centre EICV2 EICV3 EICV2 EICV3 Rwanda 43.3 68.8 95.1 59.9 Urban 39.1 72.1 52.6 34.8 Rural 44.1 68.2 103.5 64.3 Q1 33.2 52.9 96.5 69.8 Q2 37.6 61.4 95.1 66.0 Q3 45.5 69.3 102.6 63.0 Q4 47.7 74.5 98.2 61.0 Q5 51.5 84.5 85.1 44.2 Source: NISR 2012. Coping strategies 39. These findings on shocks from the quantitative data are also consistent with the qualitative understandings of that have emerged from qualitative participatory assessment (PPA) conducted in 2001 and similar ongoing social mapping and ranking exercises as part of the Ubudehe survey process (explained in more detail in 0 in Chapter 4). In the PPA, undertaking wage labour is seen as an identifying feature of households in the poorest categories, and finding wage work outside of agriculture was one of the two key areas identified as ways to escape poverty. Similarly, in qualitative research undertaken for UNICEF in 200818, households consistently mentioned agricultural wage labour as something that would be used as a coping strategy – a last resort only when other income options are not available. 18 Wylde et al (unpublished). 25 40. Other coping strategies mentioned in the qualitative work were selling of land or livestock, uprooting cassava to sell19. Borrowing from friends and family were preferred over these more detrimental strategies, however with overall high levels of poverty very few are able to provide assistance. Access to credit is very limited amongst the poor, as also documented in the CFSVA from 2009. 2.4 Conclusions on poverty and vulnerability 41. The recent EIVC III data shows that on one hand there have been some encouraging trends since the vulnerability assessment was conducted: agricultural productivity has increased over the last five years, fertilizer use has increased, and there has been an increase in the commercialisation of farming. In addition, many households have been able to diversify their sources of income, and those who have been able to take advantage of the increase in job creation (in both farm and non-farm employment) over the last five years have been able to move out of poverty most quickly. 42. On the other hand, the data above show that households who rely on farm wage labour continue to be amongst the very poorest, largely missing out on the overall trend in poverty reduction over the period. 43. Overall, the very low levels of productivity and extremely small plot sizes mean that many families are subsistence farmers, with very little surplus to use for future investment in productivity-enhancing technology or techniques. Living on the edge of survival also means that poor farmers are highly averse to risk, and therefore may not undertake investments which would be worthwhile in the long run because they could not withstand variation in returns in the short run. 19 Cassava is unusual in that it can be stored in the ground for later consumption. Uprooting it to sell rather than retain for later consumption is therefore akin to drawing down savings. 26 44. There are, however, still important gaps in knowledge that remain, mainly because information, especially on risk coping mechanisms and asset accumulation processes, is not readily available. It is hoped that on-going data analysis and collection – both qualitative and quantitative – will add to this understanding of poverty, risk, and vulnerability in order to better inform policy and programming. 27 3. Current landscape: overview of Rwanda’s social protection sector 3.1 The evolution of the social protection sector in Rwanda Social protection policy 45. The national social protection policy20 of 2005 recognises the role of risk in perpetuating poverty and the ways in which different groups are vulnerable to shocks that may compromise their welfare. It focuses on categories of vulnerable groups which are identified as being of particular importance: genocide survivors, orphans and vulnerable children, widows, prisoner’s families, demobilised soldiers, repatriates and returnees, elderly persons, victims of catastrophe, destitute people (those who cannot meet even their basic needs), and those who are historically marginalised. 46. The policy identifies three main elements of social protection:  Improvement of social status and protection of rights of the vulnerable, including legal and regulatory protection.  Protection against welfare risks through risk pooling and social insurance functions.  Transfers of income or consumption to the poor, vulnerable and marginalised. 47. Taken together, these elements provide the basis for a comprehensive system, with the better-off able to take advantage of formal social insurance mechanisms and the poor benefiting from risk-pooling arrangements (which could include microfinance) and, where necessary, direct consumption support. 48. In order to meet these aims, the policy outlines five strategic objectives for the short-, medium-, and long-term. Immediate priorities were to intensify protection 20 Republic of Rwanda, National Social protection Policy in Rwanda, November 2005. 28 measures for vulnerable groups and destitute people and to contribute to poverty and vulnerability reduction through efficient social risk management. In the medium-term the objective was to establish a coordination system for intervention in the area of social protection, and in the long-term the focus is on promoting savings initiatives and social security for all groups (the vulnerable in particular) and to promote equity and social justice through socio-economic integration and equal opportunity for all in the development process. EDPRS and the flagship Vision 2020 Umurenge Programme 49. Social protection interventions feature highly in the Economic Development and Poverty Reduction Strategy (EDPRS) of 2008-2012. The strategy has an extensive discussion of recent trends in poverty reduction, concluding that it is imperative for poverty to fall even more quickly in order to meet the MDG and Vision 2020 targets, requiring a prioritisation of efforts to increase economic growth, reduce population growth, tackle extreme poverty and food security through job creation and social protection, and ensure greater efficiency in poverty reduction through better policy implementation. 50. The conceptual framework for understanding poverty in the strategy is broadly consistent with the understanding of risk in the Social Protection Policy, but the population is identified based on their poverty status and assets – namely land and credit – rather than solely on their membership of ‘vulnerable groups’. The population is divided into the non-poor and the poor, who are then further divided according to those who have land; those who do not have (sufficient) land but who are able to work either on- or off-farm; and those who are landless and unable to work. 51. Based on these categories, interventions were to be targeted appropriately, most notably through the Vision 2020 Umurenge Programme (VUP), which is a flagship programme of the EDPRS. Under this agenda, land owners should receive interventions aimed at addressing their current risks and vulnerabilities to improve productivity and improve welfare, including activities conducted through public works employment 29 (terracing, watershed development, or irrigation) and access to microfinance and input packages. The landless who are able to work are targeted with public works employment schemes, microfinance, vocational skills, and direct support to ensure access to health and education. Those who are unable to work receive social assistance through direct support and microfinance packages that include skills development in appropriate fields. Those who are above the poverty line will be included and targeted for their potential role in employment generation, including credit packages and assistance with supply-chain management and exports. 52. The section in the EDPRS related to integrating and extending social protection is consistent with the overall strategic objectives of the 2005 Policy. It identifies several activities of importance, including:  Undertaking a feasibility study to determine vulnerability criteria and better understand implications for programme design and costing requirements.  Reviewing the legal and policy framework to simplify and harmonise the currently overlapping programmes and to improve the formal social insurance mechanisms.  Assisting different groups with the specific interventions similar to those outlined in the Vision 2020 programme.  The able-bodied poor and food-insecure households will be targeted with public works employment (cash for work), encouraged to save, and provided with vocational education and microfinance opportunities so that they can address their current vulnerability and lift themselves out of poverty.  The poor who are unable to support themselves (such as the disabled, elderly, orphans and vulnerable children, genocide survivors, and families dealing with HIV/AIDS) will be targeted with social assistance depending on their needs (some 30 will need assistance only in the short-term whereas others will need longer-term support).  In addition to this social assistance, a minimum package of services is being developed for vulnerable groups, including OVCs and families dealing with HIV/AIDS, which will ensure that they have access to schooling, health, infrastructure, as well as property rights to land and participation in governance.  Building capacity for delivery of programmes for both government and civil society.  Developing strategic plans for disaster preparedness. 53. Although a nascent sector at the time of EDPRS preparation, social protection therefore receives a very high level of prioritisation within the strategy document. Social Protection Strategy and Implementation Plan 54. More recently, building on the EDPRS, a National Social Protection Strategy (NSPS) was approved by Cabinet in January 2011. The Strategy sets out a comprehensive vision of the social protection sector over the short, medium, and long term. In so doing, it has provided a clear definition of the mission and objectives of the sector, as well as a clear definition of which activities fall within the sector21. The Implementation Plan then provides the road map for putting the Strategy into practice. 55. The mission of the sector is to “Ensure that all poor and vulnerable people are guaranteed a minimum income and access to core public services, those who can work are provided with the means of escaping poverty, and that increasing numbers of people are able to access risk sharing mechanisms that protect them from crises and shocks.” 21 Unlike other sectors such as health or education, defining the boundaries of the social protection sector is one of the greatest challenges, given that different organizations define social protection somewhat differently. The existence of the Strategy now means that the Government of Rwanda has its own clear definition on which to build. 31 The objective is therefore to: “build a social protection system that tackles poverty and inequality, enables the poor to move out of poverty, helps reduce vulnerability and protect people from shocks, helps improve health and education among all Rwandans, and contributes to economic growth.” 56. In order to achieve this objective, the social protection sector is defined as having three separate dimensions. The first is the set of ‘core’ social protection activities, which comprise the system of regular and predictable cash transfers that will provide income support to those living in poverty and vulnerable to falling into poverty as well as the expansion of social insurance (mainly pensions and health insurance) and labour standards. The second is a set of broader social protection activities that are a means of ensuring access to other public services (particularly health and education) by enabling poor households to overcome the financial barriers that they may face. The final dimension is a set of social development initiatives and complementary activities, which are focused on providing additional support to vulnerable households and helping families graduate out of poverty. 57. While this multi-dimensional definition of social protection stretches across a large number of ministries, the Strategy and Implementation Plan focus on those areas that are the responsibility of MINALOC and other core activities carried out by other ministries and agencies such as MIFOTRA and the Social Security Fund of Rwanda. Activities that fall under other sectoral policies – the ‘broad’, social development, and complementary activities – are mainly the responsibility of other ministries and sectors, although there is a role of the Social Protection sector in coordinating and monitoring these efforts. 58. There are seven outcomes in the NSPS and Implementation Plan. These are:  Leadership, co-ordination and capacity on social protection are strengthened across government. 32  Social protection policies are developed that are evidence-based and appropriate for Rwanda.  Coverage of gender-sensitive social protection programmes supporting the provision of a minimum income for families is increased.  Systems for the delivery and monitoring of social protection programmes are developed and strengthened.  Financial resources are generated and sector-wide financial systems are developed to support delivery of social protection programmes.  Social development and complementary programmes are implemented that support family and community efforts to move out of poverty; and  Awareness across Rwanda and internationally of the benefits of national social protection is increased and lessons are learned. Major programmes 59. There are a range of specific programmes already in existence that fit within the boundaries of this social protection policy. It is clear that there has been considerable progress in putting in place a safety net for some population groups, and for addressing some specific problems: The major programmes in place are:  Social Insurance: Public and formal sector workers have access to pensions, and health and accident insurance, through compulsory membership of Caisse Sociale and through health insurance schemes such as RAMA and MMI. However, although relatively wealthy self-employed people are also able to become members, coverage only extends to 4-5% of the population. 33  Vision 2020 Umurenge, which is seen as one of the three ‘flagship’ programmes of the EDPRS, is now one of the main social protection programmes, as described above. Implementation began in 30 sectors in 2008 with the public works programme, and has now scaled up to 120, also including direct support. Access to credit is also being provided by the programme, however this is not generally thought to be included as a social protection activity.  Support to vulnerable groups includes: o Needy genocide survivors are catered for by FARG, which provides support for education, health, housing, basic needs, and funding for income- generating projects. o Ubudehe (funded by the GoR and EU) provided support in the amount of 60,000 RwF to one household per village that was identified through participatory poverty assessments at the community level. o De-mobilisation of former military personnel is being addressed with support from development partners, including a $25mn Word Bank project with further support from the African Development Bank. Up to 2009/10 this was housed under MINECOFIN, but it has since been moved under MINALOC. Support includes income-generating activities, training, and cash transfers to disabled demobilised soldiers. o Refugees and returnees receive support financed by UNHCR and WFP, as well as by the Government of Rwanda through the National Refugee Council and other ad hoc transfers. 60. Related programmes in other sectors include: 34  OVCs are addressed through National Policy on Orphans and Vulnerable Children (2003), for which a strategy plan has been developed by MIGEPROF. A set of minimum standards has been established for the care and support of OVCs through the Minimum Package of Services for OVC.  Assistance with mutuelle (health insurance) membership was originally provided for roughly 10-15% of the population, though this is now changing to a graduated subsidy dependent on a household’s poverty ranking (measured by the Ubudehe category).22  MINAGRI’s One Cow One Family has been considered to be socially protective, as it provides livestock which is then expected to provide additional income- generating potential for households who can sell milk or other dairy products. 61. Over the longer term of the next ten years, the Strategy sets out a clear vision of progressively realising a ‘social protection floor’. This will involve the establishment of four key national-level social assistance instruments, which will be developed out of the programmes already existing. 22 It is important to note, however, that while in theory health insurance can be separated from other activities of the health sector, in practice it is just one part of the overall subsidisation of basic health service delivery. It should therefore really be considered as part of the wider approach to financing of the health service. 35 Table 9: Potential future social transfer programmes as part of the social protection floor Current building blocks of social protection floor Potential future social protection floor  VUP Direct Support programme Old Age Grant  FARG cash payments for elderly genocide survivors  VUP Direct Support programme Disability Grant  FARG cash payments for disabled genocide survivors  Disability payments for ex-combatants  Payments to children in VUP Direct Support Child Grant  VUP Public Works programme Public Works programme  WFP Food for Work programmes Source: National Social Protection Strategy. 3.2 Institutional arrangements 62. Social protection is a complicated sector because of its multi-sectoral nature. A broad outline of the roles and responsibilities across government are:  MINALOC: lead ministry for social protection, responsible for FARG; Ubudehe, VUP, and for setting standards and guidelines for spending on vulnerable groups at the district level. MINALOC is also responsible for demobilisation and as of 2009 for housing for informal settlements.  MIGEPROF: responsible for OVCs, child protection, and promotion of rights of women.  MINISANTE: responsible for policies relating to mutuelles de santé and financing of the mutuelle programme, as well as with basic health service delivery.  MINEDUC: responsible for policies relating to educational subsidies of poor and vulnerable and for support for special education.  MIDIMAR: A new Ministry for Disaster Relief and Refugees was created in 2011, with responsibility for early warning and disaster response and refugees and returnees (which were previously the responsibility of MINALOC). 36  MINECOFIN: responsible for oversight of the Rwanda Social Security Fund.  Districts: responsible for implementation of support to vulnerable groups including FARG and for implementation of VUP. 3.3 Expenditure patterns 63. Before looking at the aggregate view of spending patterns, we can examine some of the spending in individual programmes to gain a better sense of sources of funding and patterns of expenditure across activities. 3.3.1 VUP 64. Expenditure on VUP has increased dramatically since its beginning in 2008, as shown in the figure below. The majority of financing comes directly from Government of Rwanda (GoR) sources rather than bilateral donors. However, it is important to note that GoR spending also includes support from donors providing budget support for social protection (for example the EU and the World Bank). 37 Figure 7: VUP Revenue, by source, RwF Source: VUP Memorandum Accounts, various years. Note that revenue differs from expenditure due to the fact that the programme can carry significant balances forward from year to year, drawing down on them in later years. 65. Looking at expenditure across activities, the Public Works component is the largest activity in terms of funding, given that far more households are eligible for Public Works than Direct Support. Spending on Financial Services was large in 2009/10, the first year that the activity was added, however the net level of financing of this activity fell in 2010/11 due to the fact that some of the loans had been repaid. 38 Figure 8: VUP expenditure by activity, RwF 66. The sharp increases in spending are driven by the steady expansion of the programme as discussed in the previous chapter. Note that the increase in the 2009 Mini budget is minimal, however this was only a partial budget year, so on an annualised basis would be twice as large. 3.3.2 Social assistance: cash transfer programmes on aggregate 67. Turning our attention to spending on all the cash transfer programmes, it is clear that VUP Public Works and Direct Support drove the overall large increases in spending from 2009 onwards. 39 Table 10: Expenditure on the main cash transfer programmes, 2008-2010/11 actual, RwF million NB: VUP spending here excludes Financial Services, since this is not part of ‘core’ social protection. 3.3.3 Other social protection programmes 68. The table below shows spending on the full range of social protection programmes. Here the expansion in health insurance subsidisation and several wider social development initiatives is also striking. This includes Girinka, which increased from 500 million in 2007 to 2,800 million in 2010/11, and VUP Financial Services. 69. Unfortunately disaggregated data on demobilisation was not available, so it was not possible to separate spending on direct support for disabled former combatants from other activities for the whole group of former combatants (which include training, reintegration, and income generating activities). As a result, all demobilisation spending is included under ‘social development’ initiatives. This expenditure appe ars to be decreasing over the years, as there are fewer former combatants who are re-integrated over time. 40 Table 11: Expenditure on social protection, actual, 2004 to 2010/11, RwF million (nominal) 2004 2005 2006 2007 2008 2009 Mini 2009-2010 2010-2011 Policy leadership and management 234 472 - - - - - - MINALOC 234 472 - - - - - - Social Assistance: system of cash transfers 1,318 1,507 1,538 1,916 2,347 2,655 9,281 11,724 VUP - - - - 1,009 1,392 6,679 9,534 FARG Direct Support - 115 468 380 326 715 1,416 1,062 District earmarked transfers 834 692 1,070 1,052 1,012 549 1,186 1,128 Direct support for disabled former combatants - - - - - - - - Public works (HIMO) 485 700 - 483 - - - - Social Insurance: extension of social security 107 160 763 442 2,252 1,159 6,285 6,933 Mutuelles de sante - 95 744 430 2,214 1,139 6,143 6,914 Promotion of social security 107 65 19 12 38 20 141 18 Social development initiatives 4,112 13,558 4,927 8,150 10,422 4,633 8,810 20,269 VUP Financial Services - - - - - - 4,033 4,237 FARG Projects (IGA) 41 224 370 921 931 - 432 863 Ubudehe - 4,099 130 329 - - - - Girinka - - - 500 1,081 967 (missing) 2,800 FARG Housing 238 350 1,121 1,906 911 - 902 1,935 MINALOC - Informal settlements 313 1,218 - - 13 77 (missing) 3,132 MINALOC - support to vulnerable groups - - 12 5 393 54 242 1,701 Demobilisation (other) 3,390 7,539 3,230 4,274 6,144 2,914 (missing) 1,846 MINIYOUTH: Youth employment (IGA, Training) 69 77 - 215 860 576 (missing) 1,908 Refugees 62 51 64 - 89 45 3,201 1,846 41 2004 2005 2006 2007 2008 2009 Mini 2009-2010 2010-2011 Complementary services and activities 1,270 1,978 1,743 951 1,632 891 2,563 2,022 MIGEPROF - Child Protection 39 59 39 593 1,146 369 1,159 816 MIGEPROF - Family Protection 116 159 1,283 53 49 26 36 114 MIGEPROF - Gender empowerment 1,099 1,746 421 306 437 496 1,368 1,093 PRIMATURE - Promotion of socio-juridicial equity 16 14 - - - - - - SP Implemented by other sectors 8,517 30,776 59,818 74,313 77,372 34,019 87,606 87,057 MINISANTE - basic health care 3,926 5,261 34,048 21,111 25,783 15,488 39,769 46,709 MINEDUC - primary education 4,591 25,515 25,770 53,202 51,589 18,531 47,837 40,348 Disaster management 13 35 99 100 - - - - MINALOC / MIDIMAR 13 35 99 100 - - - - TOTAL 15,573 48,486 68,888 85,872 94,025 43,357 114,545 128,005 NB: Data for many line items is missing for 2009/2010, since the budget data available did not provide adequate detail at the sub-programme level. Spending here is largely government spending, although donor spending on VUP is also included. 42 70. Also notable is the very large increase in spending on basic services (health and primary education) over the period, reflecting government’s commitment to these areas. 3.3.4 Social protection spending on aggregate 71. Summarising these patterns into one figure below, and adjusting for inflation, we see that spending across all areas of social protection has increased by more than 13 times in real terms, over 14 times for the system of cash transfers alone and almost 17 times for basic health and education between 2004 and 2010/11. 72. It is also notable that spending on social development initiatives, including VUP Financial Services and other income generating activities accounts for a large proportion of expenditure, and has grown considerably in recent years. Figure 9: Spending on social protection (excluding basic health and education), real 2010/11 RwF million NB - 2009 Mini Budget has not been annualised. Data here includes on-budget donor spending on VUP, but excludes any off-budget donor spending on other areas. 43 73. These real increases reflect not only an increase in the total resource envelope available to government; they also reflect an increased prioritisation. The figure below indicates the share of total government spending allocated towards social protection. Both social assistance and social insurance (mainly mutuelle subsidisation) each received 1.1% of the budget in 2010/11, bringing the spending on ‘core’ social protection to 2.2% of total expenditure. Basic health and education received 10.6%. All the areas together represented over 15%. Figure 10: Government spending (actual) on social protection as a percentage of total government expenditure, 2004 – 2010/2011 74. Looking forward, the government MTEF 2011/12-2013/14 shows a continued commitment to sector financing through the main core social protection sector programs. On-budget budget allocation forecasts for the sector complemented by program financing from DFID and SIDA for the period 2011/1223 to 2013/14 are expected to increase by 13 percent from RwF48.4 billion to RwF54.8 billion. This planned 23 Figures used for 2011/12 are actual which have already been disbursed into government program account. 2012/13-2013/14 are conservative forecasts since amounts involved are commitments under effective MoUs/agreements. 44 increment might not fully cover the expansion needs but signifies a good development in line with efforts to meet the safety nets ambitious expansion plan and is consistent with the overall desired macro level fiscal sustainability. Despite the seemingly strong government commitment and sector prioritization, predictability of funds into the sector may be compromised due to lack of a strong sector MTEF that is linked to the macro- level medium term planning and reflects credibly determined sector needs. There is urgent need to strengthen the sector MTEF and its macro level linkages to enhance credible sector medium term expenditure framework planning. 3.3.5 Some observations on public financial management in social protection 75. One of the major challenges in estimating expenditure on social protection is in compiling a consistent data series on expenditure from a range of different sources. This reflects the fact that for the purposes of the Medium Term Expenditure Framework the sector is not defined in a manner that is consistent with the Strategy. Instead, the budget process unfolds in an uncoordinated way across ministries and agencies, so that MINALOC, FARG, RLDSF, RDRC, etc each embark on their own medium-term forcasting processes. 76. As a result, not only is the data inconsistent – making it extremely difficult for the sector to plan and budget – and then later monitor progress – in a coherent way, reflecting sector priorities in a way that it is clear how to hold relevant actors actors accountable for budget performance. Moving forward it will be important to ensure that the MTEF process is strengthened, with MINALOC playing the core role in leading and coordinating the planning and budgeting process for the sector as a whole. 3.4 Trends in beneficiaries 77. In line with the expenditure trends outlined above, it is also possible to trace the number of beneficiaries for the four ‘core’ transfer programmes, as shown in the figure below. Note that VUP PW and DS beneficiaries are given in terms of the number of 45 households, while RDRC and FARG DS beneficiaries are given to individuals (with the same number of households benefiting, just with benefits provided for only one person in the household). 78. The most noticeable part of the trend is a sharp decline in the number of FARG Direct Support beneficiaries from 80,000 in 2007 to 36,281 in 2008. From 2008 onwards, however, the figure on the left shows a clear marked increase in VUP beneficiary households, so that the number of Direct Support households in VUP nearly matches the number of FARG Direct Support beneficiary households by 2009/10. 79. In terms of the number of individuals reached (those living in households where at least one member is receiving one of the programmes), the picture is somewhat different. Although the number of households reached by FARG and VUP direct support was similar in 2009/10, the number of individuals reached by VUP DS is still smaller, since these households are much smaller than average (only 2.3 members compared to 4.9 members in FARG households)24. However VUP will continue to scale up each year, so the number of individuals reached by VUP DS would be expected to catch up with FARG DS fairly soon. 24 These assumptions are based on the following data sources: For VUP DS the VUP Annual Report 2010 was used; for FARG DS the Census of Genocide Survivors from 2007 was used; for VUP PW the VUP Survey Report from 2009 was used; and for RDRC the average household size nation-wide as assumed, in the absence of more specific data. 46 Figure 11: Beneficiary numbers of ‘core’ social protection programmes, 2007 – 2009/10 80. In 2009/10 in total it is estimated that around 475,000 individuals were reached by these social assistance programmes combined, or roughly 4% of the total population. This is of course far below the number of extreme poor (24%), however this is largely a result of the fact that VUP has yet to roll out to the entire country. In 2009 it was estimated that in the sectors in which VUP had reached, 5% of individuals were covered by Direct Support and 14% by Public Works for a total of around 19%25. 3.5 Summary of key existing issues and challenges and the implications for the assessment 81. Bearing the above context and objectives in mind, the remainder of the Assessment aims to build on the existing knowledge of the sector to identify the key issues that will be faced in putting the strategy and implementation plan into practice. The Assessment is therefore not a general survey of the sector, but rather a very targeted one. This section briefly outlines what is already known about some of the key issues, as this existing understanding will inform the approach and methodology employed by the Assessment. 25 As per the VUP Survey Report (2009). 47 Decentralisation: implications for social protection 82. Rwanda’s decentralisation process has been ongoing since 2000, and has now entered its third phase with the new Fiscal Decentralisation Strategy (2011-2015). The third phase focuses on the need to consolidate the reforms undertaken so far, improving implementation of the policy rather than setting new objectives. The major challenges so far were identified as resources not matching devolved roles and responsibilities; earmarked transfer mechanisms and procedures that constrain LG functioning; responsibilities that overload district staff and general capacity constraints both in terms of human resources and systems. The strategy therefore articulates an objective of reducing the extent of earmarking (or tying) of funds and allowing more discretion at district level. That being said, sectoral ministries are still responsible for the “establishment of sectoral policies, strategies and guidelines for local service delivery as well as the provision of fiscal transfers to enable services delivery, and monitoring of budget execution and performance in their respective sectors”. 83. While recognising that matching funds, functions, and functionaries is one of the basic requirements for efficient decentralisation in general, decentralisation often involves a careful balancing act of allowing greater discretion to improve service delivery while at the same time ensuring that the centre maintains enough scope to effectively manage results in terms of equity and quality. Even in developed countries, social policy is generally shared (rather than fully devolved); central-level politicians are reluctant to fully hand over responsibility to lower levels, since it is the centre that is ultimately held accountable for results26. This is especially true for aspects of service delivery that are entitlements (where citizen’s rights are enshrined in legislation) as opposed to those areas that are discretionary programmes. 84. Because of the explicit focus on equity and the redistributive nature of social protection, in most countries it tends to be even less decentralised than, for example, 26 OECD (1997) “Managing Across Levels of Government”. Paris: OECD. 48 education. The Decentralisation Strategy’s emphasis on reducing earmarks on spending will therefore have a potentially large impact on social protection policy, as it will require that districts adequately prioritise social protection activities if given the choice. There is an urgent need to better understand the implications of the decentralisation strategy for the sector, and for MINALOC to proactively and strongly engage with MINECOFIN on implementation of the proposed changes. MINALOC social protection spending: need for more strategic focus 85. These questions around decentralisation as it relates to social protection must ultimately boil down to the policy objectives underpinning the spending line items and the functions that have been given to districts in reaching these objectives. At present, there is some scope for articulating these objectives more clearly especially for MINALOC’s social protection activities outside of VUP. 86. The Social Protection PER27 found that although MINALOC does participate in the budget process, including the drafting of a detailed MTEF proposal through a lengthy internal process whereby ministerial priorities are articulated and proposals are justified, in practice MINALOC’s three social protection sub-programmes28 (Promotion of social security, support to vulnerable groups, and support to vulnerable groups self- promotion) remain largely ad hoc, in terms of both budget allocation and execution. In practice, these line items tend to serve as a reserve for other activities, rather than as strategic budget priorities in their own right. Much of the spending was provided to districts, which had submitted requests to fund vulnerable groups (returnees, historically marginalised, etc.). This process appears to be fairly discretionary, without any allocation criteria, which probably reduces equity as well as effectiveness of spending. Based on discussions during the preparation of the PER, there needs to be more oversight within MINALOC on the strategic nature of spending on the three sub programmes, even if there are formal accounting procedures in place. It was not 27 Government of Rwanda (2010). 28 These three sub-programmes are the main channels for MINALOC’s social protection spending outside of VUP. 49 apparent who was responsible for spending on the three social protection sub- programmes outside of the Directorate of Finance, suggesting the need for more focus on results in implementation. 87. MINALOC has already made some progress in streamlining these funding lines, eliminating the three sub-programmes at central level and replacing them at district level with one earmarked transfer (“support to Vulnerable Groups”). MINALOC is in the process of defining the guidelines for districts more clearly, and this is an opportunity to ensure a more strategic approach is taken with this funding line. 88. A wider list of bottlenecks identified by the PER are shown in the figure below. Figure 12: PFM bottlenecks across the policy and budget cycle: impact on social protection Source: Social Protection Public Expenditure Review. 50 Fragmentation of activities 89. Of these other major issues in the sector, one of the most important is that there are multiple fragmented processes for beneficiary selection and identification, delivery of benefits, and reporting mechanisms across the various programmes. This issue requires immediate attention, as already identified by the strategic objective in the implementation plan of consolidating parts of FARG and demobilisation activities under VUP. 90. Furthermore, districts are currently responsible for implementing social protection activities, yet it is not always clear:  How guidelines (where they exist29) are implemented in practice.  Whether district administration officials have enough capacity (including staff time) to do so.  Who is responsible for what kinds of monitoring and reporting; and  To what extent results at district level are managed by central ministries and agencies. Implications for the SSN Assessment 91. Unravelling these different issues within the sector will involve, firstly, mapping out in more detail how programmes are implemented in terms of both programme design (how things should work in terms of targeting, payments, reporting, budgeting, etc.) and actual practice on the ground. This mapping will hopefully then allow a clearer view of the potential options available for harmonising programmes and problems that are likely to arise. The Assessment therefore aims to provide a tool for stakeholders to 29 VUP has clear operating guidelines at district level that were developed at the programme’s outset (though practices may have evolved since then); MINECOFIN released guidelines for decentralised line items; and other guidelines may exist but were not able to be sourced by the team. 51 work through the various options, assess their feasibility and implications, and ultimately make a recommendation as to the best way forward. The scope of the assessment therefore includes both issues around the design of programmes (consistency and effectiveness of objectives, target groups, benefit packages, etc.) as well as their implementation and financing. 52 4. Systems review: from programme design to implementation 92. This chapter provides a systematic look at design elements (such as objectives, targeting approaches, benefit types, transfer values, and delivery methods); implementation mechanisms; accountability mechanisms (routine monitoring and reporting); and evaluation approaches of the main social protection programmes. While some of these aspects have been well documented in other reports, information tends to be scattered across a range of disparate sources and is not easily compared and analysed. 93. Chapter 3 above noted that as the sector begins to move forward with the implementation plan, there is a need to move from generalities (such as the recognition of the need to consolidate, harmonise, and coordinate actions) to specifics (where exactly are overlaps across programmes? Which implementation arrangements would need to change? Which beneficiaries would be affected?). This chapter therefore provides the detail that will be necessary for the subsequent analysis, conclusions, and ultimately recommendations later in the report. It covers both ‘core’ social protection programmes as well as selected complementary programmes that are relevant for the assessment, as discussed in the introduction. Box 1: A note on data sources for the systems assessment As is noted throughout this report, there is a lack of comprehensive and systematic evaluation data for the social protection sector as a whole. As a result, the systems assessment in this chapter as well as the analysis throughout the document pulls together the scant information that is available from a set of disparate programme documents, evaluation reports for specific programmes, and joint sector review reporting. Some of the more relevant evaluation documents for VUP include: • 2010 and 2011 DFID evaluations of VUP (see Devereux 2010 and 2011). These are based on a review of documents and stakeholder consultations, but do not include primary data collection. • The 2009-2010 VUP Annual Report. • VUP Poverty Survey. This involved primary data collection from a household survey, but there are some methodological issues which are discussed in further detail below. The only primary survey data available is therefore from the Poverty Survey, annual Targeting Surveys, and the Baseline survey. 53 4.1 Social Assistance: system of cash transfers30 94. The core social protection programmes that are outlined in turn include:  VUP Direct Support.  VUP Public Works.  FARG Direct Support.  Support to disabled demobilised soldiers; and  MINALOC decentralised funding. 4.1.1 The Vision 2020 Umurenge Programme (VUP)31 Origins 95. VUP emerged from a high-level leadership retreat in February 2007, as a response to what was felt to be slow progress on the government’s poverty reduction agenda, in spite of positive economic performance. The programme is therefore government-owned and led, even if it receives donor support (namely from DFID, the World Bank, the EU, UNICEF, and Sida). The overall goal of VUP is to contribute to the national target of reducing extreme poverty from 36.9% in 2005/6 to 24% in 2012, by accelerating the reduction of poverty in those areas where the programme is implemented. Programme components 96. There are three components to VUP (which will be explored in more detail separately below): 30 Regarding terminology, the SP Strategy categorizes both the system of cash transfers and the extension of social security and labor standards as being part of core social protection. In the international literature, these are commonly referred to as social assistance and social insurance respectively. We therefore use them inter-changeably in the report. 31 Information on VUP is drawn from: MINALOC (2011) Vision 2020 Umurenge Programme Annual Report 2009/10. 54  Direct support: unconditional cash transfers.  Public works: paid employment on productive community asset development projects; and  Financial services: increasing access to financial services, including microcredit, a matching-grant challenge fund and financial literacy (further details provided in section 4.3.1 below, since this component is not considered part of ‘core’ social protection). 97. The rationale is that by including three different components, VUP will be able to appropriately address poverty of different groups of the poor: those who do not have labour capacity such as the elderly, disabled, or child-headed households who can benefit from direct support; those who do have labour capacity and can benefit from public works employment; and those who are able to take out a loan or participate in an income-generating activity. Geographical and programme coverage 98. The programme began in 30 pilot sectors (the poorest sector in each of the 30 districts) in May 2008, beginning with the public works component, and then the direct support component was launched in January 2009. In 2009/10 the programme expanded to include a further sector in each district (reaching 60 in total) with another 30 sectors added in each fiscal year so that in 2011/2012 there are now a total of 120. Financial service was launched in January 2010 (the Ubudehe Credit Scheme) and is now in all 120 sectors. The scale-up plan envisages that all sectors in the country will be covered for Direct Support by 2016, and by that time 240 sectors will have been reached for Public Works. Targeting and graduation 99. VUP programme eligibility is based on the Ubudehe targeting approach (see Box below). Starting in 2009, a targeting exercise should be undertaken annually, 55 determining the eligibility of a household for the three programmes based on their Ubudehe category and the availability of working adults:  Households in Ubudehe categories 1 and 2 with no adult labour capacity (meaning adults 18-65 capable of manual work32) are eligible for direct support.  Households in Ubudehe categories 1 and 2 with adult labour capacity are eligible for public works employment; and  Households in Ubudehe categories 1 through 4 are eligible for financial services of different types subject to fulfilment of additional loan specific requirements. 100. Prior to the 2011-2012 fiscal year, an additional criteria based on land ownership was included, so that in order to be eligible for either direct support or public works a household could not own more than 0.25 hectares. This criteria was dropped because it was felt that the land criteria was too crude (in particular not taking into consideration the quality of the land, or the ability of a household to cultivate it if lacking in labour capacity) and was largely irrelevant for sectors in Kigali. 101. It is envisaged that households should ‘graduate’ from direct support to public works to financial services, thereby providing a ‘staircase’ of support and out of dependency. In practice, however, households eligible for direct support are likely to remain eligible over time, as those without labour capacity (particularly households with elderly and disabled members) are likely to need ongoing support. 32 Reasons for an inability to work include: old age (65 and older), illness, disability, or being a student. 56 Box 2: Ubudehe targeting Although participatory poverty assessments (PPAs) generally allow individual communities to articulate their own definitions of poverty, over the course of the early PPA exercises in Rwanda it was found that definitions tended to be similar across communities. As a result, the Ubudehe targeting criteria have been globally defined in order to allow greater consistency in application. The categories are:  Category 1: Umutindi nyakujya (those in abject poverty). Those who need to beg to survive, without land or livestock, or adequate shelter, clothes, or food. Children are malnourished as a result, and these households fall sick often without access to medical care.  Category 2: Umutindi (the very poor). Characteristics are similar to Category 1, but this group is physically capable of working on land owned by others even if they have either no land or insufficient land to produce enough for survival needs.  Category 3: Umukene (the poor). These households have some land and housing, and can live off their own labour and production. They have no savings, and while they may have enough food it is not very nutritious. They have no surplus to market. Children do not always go to school.  Category 4: Umukene wifashije (the resourceful poor). Similar to Category 3, but with some small savings or assets, and children attend school regularly.  Category 5: Umukungu (the food rich). This group has larger landholdings and livestock, and often have paid jobs.  Category 6: Umukire (the money rich). This group is similar to Category 5, but often have salaried jobs, good housing, and sometimes a vehicle. They have enough money to lend and to get credit from a bank. During the participatory targeting process, every household in the country is given an Ubudehe category for that year, based on the assessment of the household’s well being made by the community. 102. In the initial stages of the programme in 2008 and the 2009 mini-budget year33, the process was explained during a community meeting. There was no specific targeting list form used. 33 In mid-2009 Rwanda harmonized its fiscal year with the East African Community, moving from a January-December year to a July-June year. In order to make this switch, in 2009 there was a ‘mini’ (half- year) budget from January to June, with the new fiscal year system beginning with 2009/10 (from July 2009 to June 2010). 57 103. In order to provide greater clarity, transparency, and consistency and reduce the scope for debate over criteria, in 2009 the process was strengthened through the introduction of a Targeting Survey, which is essentially a census of all households using a targeting form collecting basic data on eligibility34. Training was also carried out to ensure that the supervisors of community targeting meetings were clear on how to use the form. 104. The Targeting Form is then validated by the Executive Secretaries at sector level as well as district Vice Mayors in charge of Social Affairs. Budget allocation 105. Initially, funding was meant to be allocated across the three components according to national guidelines of 20% for Direct Support, 50% for Public Works, and 30% for Financial Services. However, in practice these varied based on the number of eligible beneficiaries in each sector. 106. According to new guidelines, there will be a two-staged budget allocation:  Geographic allocations to sectors based on the estimated needs for direct support, public works, and financial service needs. In areas where the programme is already running, this estimate will be based on the previous year’s beneficiary numbers, since Ubudehe status is unlikely to change drastically from one year to the next.  Within sectors, the allocation across the three programmes is as follows: 34 The fields include: name, date of birth and ID number (of household head), number of adults, household size, number of adults unable to work and, if any are unable to work, the reason why. 58 o Direct support is viewed as an ‘entitlement’, meaning that all households that are eligible must receive a transfer. Direct support therefore is allocated as much of the budget as is necessary to reach all beneficiaries. o The remaining balance is allocated across public works and financial services. There is a cap per sector on the transfers for financial services, so once a sector has reached that cap then it will no longer receive further transfers for that function (since ongoing loans should then be fully financed from loan repayments rather than new injections of funding). Programme management 107. VUP initially sat within the Community Development Fund (CDF) in MINALOC, and now CDF has become the Rwanda Local Development Support Fund (RLDSF). While it was part of CDF, the VUP Management Team included Rwandan staff and international technical assistance covering areas of programme development, financial management, and monitoring and evaluation. Under the new RLDSF structure, some of these functions will now be transferred out of VUP into cross-cutting RLDSF units (for example, the RLDSF M&E unit will now be responsible for monitoring and evaluation of all RLDSF programmes including VUP). 108. At the implementation level of sectors, the programme has a Finance Officer and VUP Program Manager, and there is one M & E officer in each district. 109. With respect to financial flows, through the 2010-2011 fiscal year transfers were made through the VUP programme account that pools funds from government and donors. Sectors would request disbursements from the VUP account. For Direct Support, the disbursements should have been requested quarterly and then the sector should have made monthly payments into beneficiary accounts. For Public Works, disbursements should have been requested according to the cycle of a given public works project (30% at the outset; 40% and 40% subsequently). Sectors were then to pay beneficiaries based on the number of days worked. 59 110. Some initial challenges in the financial management of the programme during the first three years were that:  There was often a lack of timely requests for disbursements from sectors, which often led to delays in payments to beneficiaries.  Delays in the targeting exercise severely delay payments at the start of the fiscal year; and  Sectors often faced delays related to the procurement of public works, particularly if they were waiting for their first disbursements following the (delayed) targeting exercise. 111. As of the current fiscal year (2011-2012), payments are now ‘decentralised’ meaning that the VUP headquarters no longer plays an intermediary role in the flow of funds. The idea is that by removing this stage of the expenditure process there will be greater speed in getting payments to beneficiaries. This will need to be carefully monitored, however, because, conversely, this also means that VUP headquarters will no longer be able to follow up with sectors which are lagging behind in requesting disbursements. 112. In addition, sectors are being encouraged to ensure that more advance planning – including the launch of procurement processes – is done in advance of a new fiscal year so that public works projects can continue without delay even in the first financial quarter. Monitoring and evaluation 113. There are a range of routine monitoring processes in place including:  Financial reporting. 60  Non-financial reporting.  Joint donor/government reviews. 114. The monitoring system has continued to evolve as the programme has expanded. A comprehensive list of indicators – used by both donors and government – is in place to measure progress in implementation. Monitoring tools have been developed to collect the required data, which include monitoring of payments to beneficiaries, financial reporting, measuring progress on construction of public works assets, and monitoring of market wage rates. These tools have been improved over time, but until now have remained fairly manual. A comprehensive MIS system is now being put in place to ensure more accurate, timely, and seamless reporting for performance management. 115. What is still lacking, however, is a comprehensive evaluation framework. Due to the fact that the initial focus was on piloting and rolling out service delivery, rather than spending a long period in design, and due to the choice of targeting the poorest sectors in each district first, a proper controlled evaluation strategy has been a challenge. A baseline survey was undertaken, with the aim of providing the basis for an experimental evaluation, however the baseline is considered to be somewhat methodologically problematic (see box below). 116. The focus for the measurement of impacts has instead shifted to the Poverty Survey undertaken in 2009, which largely relies changes in Ubudehe status to assess ‘graduation’ and therefore success of the programme. This survey also suffers from some methodological weaknesses, including in particular the lack of a control group against which to compare results and some weaknesses in the collection of poverty data. The desire to provide an evaluation tool that is more ‘decentralised’ and therefore easier for districts to collect data appears to have taken precedence over obtaining robust and credible evaluation data. 117. As a result, the programme is not currently able to provide any real evidence on impacts and outcomes of the programme. To remedy this situation, some proposals for 61 a robust randomised control approach have been submitted to a third-party funding organisation (3iE), and it is hoped that these will be accepted and the current gap in evaluation will be closed. Having measurable results will be especially important as the programme continues to be scaled up across the country, as the Government will need to base further policy decisions on solid evidence on the value for money in social protection programmes. Box 3: Methodological issues with the Baseline and Poverty Surveys Baseline survey This does include both treatment and control location. However the stratification approach was based on a two-stage stratified sample, where sectors were selected in the first stage based on the number of poor households as determined by MINALOC. Unfortunately, the listing of poor households was not provided to the survey team, and the number of poor households found in the treatment and control districts varied significantly from the initial estimates. As a result, the sample will produce less precise estimates. Even aside from these issues with precision, there is an inherent problem with using a qualitative assessment of the number of poor households as a basis for selecting sectors to include in the programme and the control, since one of the key research questions is indeed to test the robustness of this initial selection of districts. Interestingly, the Baseline Survey results indicate that the control areas had higher levels of poverty than the treatment areas. Poverty survey The poverty survey is problematic both in the methodology used, as well as the interpretation. In terms of the former, the key issues are: • The sampling strategy was based around Ubudehe status, with an over-sampling of Ubudehe categories 1, 2, and 3, and much less sampling of the better-off categories. While this makes sense in some ways, given that those in lower Ubudehe categories are the ones eligible for the programme, it is problematic from the perspective of being able to assess the extent to which Ubudehe categories themselves accurately reflect poverty as commonly measured by consumption. It makes it very difficult to assess the extent of errors of exclusion, which should be just as important as errors of inclusion. • The consumption measure used is problematic. In order to save on survey costs, a much shorter list of consumption items is included. Given the importance of assessing Ubudehe categories against a robust consumption measure, this decision might not have represented the best trade-off in terms of balancing costs against the quality of the evaluation. It might have been better to include a consumption module that was consistent with the EICV so that results could be triangulated. In terms of interpretation, the report also draws some conclusions that are not entirely appropriate. For example, it uses changes in Ubudehe status as an indicator of ‘net graduation’, so that if a household is deemed to be in a higher Ubudehe category from one survey round to the next it is seen as having ‘graduated’ out of poverty. It is important to always remember, however, that Ubudehe categories are still subjective measures, and just because a household has been allocated into a higher category does not necessarily mean that it is significantly better off. Of course, the subjective measure is an interesting data point, but it should not be used to draw any conclusions on the impact of the programme. Additionally, the survey report interprets the fact that Direct Support disproportionately benefits older people whereas Public Works benefits children more as being a problematic result. The fact that children are not as well represented under Direct Support is simply the consequence of the design of the programme, since children disproportionately live in households with able working-age adults. By no means does this mean that Direct Support is ineffective in its targeting design; on the contrary, it would indicate that intended beneficiaries (who are likely to be disproportionately elderly households who are unable to work) are being included under that arm of the programme. 62 Programme-wide results in practice Targeting 118. The Targeting Survey revealed a lower level of direct support eligibility found in practice (3% of households instead of 5% suggested by the Poverty Survey), which suggested that some eligible households are either not aware that they are entitled to a transfer or are wrongly assigned to public works employment when adults are recorded as being fit for work even if they are not. To address the former, it was proposed that a beneficiary ID card is provided, clearly stating the entitlement to direct support. To address the latter, changes were made to the targeting form to simplify the assessment of whether adults were able to work or not, and to implement an appeals process so that those people who believe they were misclassified can have their case reviewed. The pilot of the appeals process indeed found that around 40% of appeals were from people who were marked as being fit for work but who suffered from chronic illness (such as AIDS) who were seen to be ‘fit’ by their community or who were uncomfortable with revealing their health status in the public Ubudehe targeting meeting. 119. Starting this year, the Ubudehe category allocated to households will determine not only eligibility for VUP but also the ability to access free and subsidised health insurance. As the population starts to realise that being allocated into Ubudehe category 1 or 2 carries with it significant benefits, more and more households may advocate for their inclusion into these categories, with implications also for VUP eligibility. Red Flag: Risks to targeting efficiency It will be very important to closely monitor the targeting of households over the next few targeting cycles to see if there is an inflation in the numbers of households falling into categories 1 and 2. There is a risk that as more and more programmes start to use Ubudehe for targeting, the system may begin to break down as households learn to ‘game the system’. 63 VUP Direct Support There were 9,962 households receiving Direct Support in the fiscal year 2009/2010. Transfer size and predictability 120. Transfers depend on the size of the household, at a rate of RwF7,500 per month for the first person and increasing to a maximum of RwF21,000 for a household of 5 or more people. 121. Direct Support transfers are seen as an entitlement for households that meet the eligibility criteria, meaning that all eligible beneficiaries receive transfers (i.e. there is no rationing). 122. The first Annual Review in 2009 found that payments were not being made in a predictable manner. While there has been some improvement, the problem has continued for Direct Support. Transfers should be made on a monthly basis, but almost all sectors pay them on a bi-monthly basis instead. This is because they receive transfers into the sector VUP account every two months, and instead of transferring the amount on a monthly basis they simply transfer the full amount each time they receive a disbursement. Delays are also exacerbated by a lack of timely requests for disbursements from the sectors as well as slow payments to sectors from the centre. 64 Red Flag: Risks to programme effectiveness in poverty reduction The delays involved in the expenditure process (for all the reasons listed above) are likely to have a major impact on the effectiveness of the programme. For Direct Support, evidence from other countries has shown that effectiveness will depend largely on the reliability and predictability of the transfers. So efforts should be made to ensure that households receive the transfers every month (not every two months as currently happens) and that there is no gap in the payment due to delays in annual targeting processes. For public works, delays due to procurement reduce the number of days of work provided each year and therefore limit the impact on households. It will be important to monitor this closely now that funds have been decentralised and no longer flow through RLDSF as an intermediary. Targeting 123. The evidence from case studies and the Targeting Survey showed that, unsurprisingly, Direct Support largely benefits households with elderly members who lack sufficient means to take care of themselves. In this way, Direct Support is functioning partially as an old age grant, albeit one that only selects those households with Ubudehe categories 1 or 2. The Targeting Survey argued that this was evidence of ‘bias’ against children in the programme, however this statement is perhaps not entirely accurate. Given that it is specifically targeted at households without labour capacity, children are simply far more likely to live in households benefitting from public works than direct support. 124. The Targeting Survey argued that women with young children should be classified as ‘not available for work’ so that single female households with young children could be eligible for Direct Support rather than public works, and that this would contribute to enhanced child nutrition outcomes as a result. However, it is important to note that the number of households meeting this expanded criteria is likely to be fairly small. It is also important to ensure that this would not provide perverse incentives for males to leave households with young children in order to make the household eligible for the transfer. 65 125. Ultimately the Direct Support transfer as currently designed is simply not oriented towards achieving child welfare outcomes (as would be the case with a specific child grant); rather it is more akin to a disability/old-age grant. This does not necessarily reflect a deficiency with the Direct Support programme, rather more a reflection of different objectives. Of course, there will be many children who will be expected to benefit from Direct Support where they are living with eligible beneficiaries. However, if child nutrition outcomes more generally are a specific objective, then a different programme design – one that explicitly targets households with children 0-2 - is probably called for. VUP Public Works Targeting 126. There was significant leakage to non-eligible beneficiaries in the 2009/2010 fiscal, year, since 22,270 work opportunities out of the total 61,355 created went to households not in Ubudehe categories 1 or 2. This was because there was a greater emphasis on the timely completion of certain projects (and therefore a need to bring in more labour than the number of eligible beneficiaries) over the objectives of providing the maximum amount of labour for eligible households. 127. At the same time, there is also a significant degree of job rationing due to the unavailability of places. The 2010 DFID Review (Devereux 2011: 7) found that in general the Public Works component was performing more as an asset creation programme – achieving objectives in infrastructure and environmental conservation – than a social protection programme. According to qualitative discussions with programme managers, the objectives were centred around the productive outputs rather than providing employment as a social protection instrument. 66 Red Flag: Risks to programme effectiveness The public works component is not delivering on the commitments to provide six months of employment to beneficiary households. Instead, the average number of days of employment was 70 (instead of 120), leading to a significant erosion in the value of the transfer. The average earnings were only RwF63,000 in 2009/10 compared to RwF169,000 for Direct Support. As a result, the programme is unlikely to have any significant impact on poverty reduction. Programme costs and efficiency 128. Wages are set locally by VUP staff, sector officials, and the community to ensure that they are consistent with the prevailing market wages for daily casual labour. A comparison of local wage rates with the Public Works wages in 2009 found that in 14 out of 30 sectors wages were actually higher than the market rate by about 10% on average. This led the Technical Audit to recommend that rather than continuing with this local determination of the rate, the setting of wage rates should be centralised. The 2010 DFID review disagreed with this recommendation, on the basis that prevailing market wage rates tend to be unreasonably low because there is far more supply of labour than there is demand. As a result, allowing the public works programme to exert higher pressure on daily wage rates would be a positive indirect benefit of the programme. 129. A further challenge with respect to the wage rate is that there has been a significant degree of wage inflation, so that on average the local average wage rate increased by 30% between November 2009 and July 2010 (Devereux 2011: 19). While the linkage of VUP wage rates to market rates ensures that the value of the transfer is not eroded by inflation, in the face of fixed budgets for public works, increasing wage rates would effectively lower the number of available work places the programme could offer (unless these wage increases are explicitly worked into budget increases). The result is therefore a trade-off between the value of the transfer and the number of beneficiaries. 67 Work in Progress: Programme efficiency So far, the percentage of expenditure going towards the wage bill has been very high (at around 88% since the programme began), which does at least ensure that as many resources as possible reach beneficiaries. However, this high ratio is unlikely to be sustained, since it resulted from the fact that most early projects have been focused on terracing – work that requires few inputs – but these are becoming exhausted. Going forward other types of projects will need to be implemented instead, with a lower percentage going towards wages and also more difficult project procurement and management. 4.1.2 FARG Direct Support 130. The Fond d’Assistance aux Rescapées du Genocide (FARG) was established by law in 1998, with the objective of collecting funds and distributing benefits to survivors of the genocide. The number of beneficiaries increased from around 23,000 in 2001 to a high of 80,000 in 2007, although numbers have dropped significantly since then, falling back to 23,000 in 2010. This increase then decrease is likely to be at least partially driven by the demographic profile of beneficiaries indicated below; as the cohort of orphans going to school is now finishing it would be expected that numbers of eligible beneficiaries would decrease. Targeting 131. According to the guidelines, eligibility for Direct Support transfers is based on the following criteria:  Orphans under the age of 21 that have been adopted by other families who have no property that provide them with incomes as a source of the upkeep.  Double orphans who survived the genocide and currently attend school with nowhere to stay during the school holidays. 68  Double orphans who survived the genocide staying in a poor family that is unable to meet the basic necessities of life.  An old-aged survivor of the genocide who is unable to work and cannot take care of him - or herself.  A survivor of the genocide with a physical disability or chronic disease that prevents him/her from working and who is without any source of income. 132. Beneficiary selection and identification is undertaken at the community level, facilitated by the Sector Executive Secretary. A list is drawn up of ‘needy’ genocide survivors who fit the criteria above. This list is then validated by the Sector Executive Secretary and by the district. Transfers 133. Unlike VUP, transfers are made to individuals rather than households. The amount is lower than VUP, set at RwF5,000 per month. Monitoring and evaluation 134. The routine monitoring of FARG has been problematic. The MIS system initially did not allow the programme to track the number of unique beneficiaries over time, as it would write over the record for prior beneficiaries when new beneficiaries were added. Work in Progress: Programme effectiveness A new MIS system is currently being designed for FARG to overcome some of the weaknesses in the past. This work will also be coordinated with the wider sector- wide development of a coherent and consolidated MIS system which is also currently under review. 69 4.1.3 Disabled demobilised soldiers 135. In 2007 a law was established relating to the protection of disabled former combatants. Disability status is established through a medical screening process, and individuals are assigned to one of four categories based on the degree of disability (with category 1 having the highest). 136. The RDRC provides social assistance to these disabled former combatants in the form of a subsistence allowance, along with other benefits (which are not part of ‘core’ social protection, such as housing support, income generating activities, medical insurance, etc.). The subsistence allowances are significantly larger in value than either FARG or VUP direct support: RwF50,000 per month for those with a disability category of 1; RwF35,000 for those in category 2; RwF25,000 for category 3 and RwF20,000 for those in category 4. However, the total number of beneficiaries is very small: 193 in category 1;409 in category 2; 525 in category 3 and 1,407 in category 4, for a total of 2,534. 137. Eligibility for the transfers is based on the level of vulnerability of the disabled former combatants in addition to the medical screening. The latter is conducted for all beneficiaries every two years by a medical screening committee. Vulnerability is determined by sector Executive Secretaries in charge of Social Affairs, who are given a set of guidelines. 138. Transfers are then made to individual beneficiary bank accounts. 4.1.4 MINALOC decentralised funding 139. A small level of earmarked transfers is made to districts for the purposes of social protection. Prior to the current financial year, there were no guidelines attached to this transfer, so districts decided themselves how to allocate the funds. The value of the transfer is very small, so in practice districts tended to use it as a small fund for 70 providing one-off assistance to households needing support. There was no formal ‘eligibility’ criteria; rather, the needy would simply approach the district with sm all requests and the district would prioritise as best they could. Given the ad hoc nature of the fund and the very small size of transfers being made, districts did not tend to monitor or even record the transfers, so there was virtually no monitoring – let alone evaluation – of this spending. 140. With the changes to decentralised transfers in the current fiscal year, MINALOC has now produced a set of guidelines for the use of the transfers. Households are now to be selected based on Ubudehe categories (1 and 2), and are to be used for income generating activities. Work in Progress: Strategic orientation of the sector As will be discussed in more detail in Chapter 5, the new guidelines for decentralised transfers are somewhat problematic in that they seem to shift the nature of the transfer, making it now a ‘complementary’ programme rather than consistent with core social protection. 4.2 Social insurance: extension of contributory social security and labour standards 4.2.1 Expanding coverage of formal social insurance 141. Currently the only government programming on expanding coverage of social insurance is a small line item under MINALOC for policy work in the central ministry. Although the Social Security Fund of Rwanda (SSFR) reports to MINECOFIN, this is a largely regulatory role. MINECOFIN has a duty to ensure that benefit packages are appropriate and that the SSFR is financially viable, and according to law the Government serves as a guarantor for the Fund. It does not, however, currently provide any direct public funding for the Fund35. 35 The GoR does pay pension contributions for its staff – as do all formal employers – however this is better understood as part of staff remuneration rather than an active social protection intervention. 71 142. The 2009 Social Security Policy identifies a number of weaknesses in the current social security system. These include:  Low levels of coverage: currently only around 7% of the population was in the formal sector, where social security contributions are mandatory. The rest of the population in the informal sector is left out of the current system.  Low levels of benefits: The current system is based on defined benefits (as opposed to defined contributions), with the minimum level defined at 50% of the minimum wage. The monthly minimum paid is only adjusted on an ad hoc basis by Presidential decree, which has led to substantial erosion in the value over time due to inflation. The average value of the pension at the time of the 2009 review was only RwF3,500 per month – not substantially different from the transfers made under VUP in spite of the fact that it is a contributory system.  Inequities in financing: At the same time, there is no ceiling on benefits paid and large disparities across members in terms of salaries, which leads to the situation where contributors with small salaries fund some large pensions for those who had high-paying jobs before retirement.  Unsustainability of financing: the current defined benefits system is only sustainable, according to recent actuarial estimates, when there are ten contributing workers for every retiree – a situation that will not be sustained for long due to demographic changes. Investment returns have been low so far.  Fragmented management: until the time of the policy in 2009, social security was managed by a range of different institutions: the Social Security Fund (SSFR, previously the Caisse Sociale) for pensions, RAMA (La Rwandaise d’Assurance Maladie), MMI (Military Medical Insurance), and mutuelles de santé for health care. RAMA and SSFR have now been merged into the Rwanda Social Security 72 Board (RSSB), with the expectation that MMI and the mutuelles will eventually be integrated as well. 143. In order to address these concerns stipulates that by 2020 every resident in Rwanda must have some kind of pension cover, either public or private. Those workers in the informal sector who are part of a cooperative will be expected to join into pension schemes, with an objective of reaching at least 70% coverage (on the basis that the remaining 30% are below the poverty line and would not be able to afford any kind of pension contribution). According to a recent interview with the RSSB, implementation has already resulted in an increase in coverage from 7 to 18% in just one year, and contributions have increased by 35%36. 144. As the pension branch of the social security system is reformed, it is expected that the current ‘defined benefits’ scheme under SSFR will be supplemented with a Provident Fund (the second ‘pillar’) that will be a voluntary contribution based on defined contributions. In order to incentivise the expansion of this second pillar the government expects to provide public subsidies, mainly through tax exemptions. It is expected that this subsidy will have positive benefits for the expansion of social security and investment, economic growth, and eventually even overall tax revenue. 145. Further studies are expected to be undertaken to assess the costs and benefits of these envisaged reforms. 4.2.2 Health insurance: the mutuelle de santé 146. Rwanda’s health insurance, the mutuelle, has been noted internationally for its success, increasing coverage rapidly from about 7% in 2003 to over 90% in 2010 (GoR 2011: 73), in line with the policy decision to make health insurance coverage (whether through formal employer-based insurance or the mutuelle). The Community Based Health Insurance (CBHI) Policy of 2004 adopted an approach of charging a flat rate for 36 ISSA 2011. 73 the insurance, in order to simplify communications messages and encourage take-up. Households were initially charged one flat rate regardless of size, and then this was adjusted to charge RwF1,000 per household member. Subsidies funded by the Government and many donors – most importantly the Global Fund – were provided for ‘indigents’, roughly 10-15% of the population. 147. In addition to the premium, households pay 10% of the cost of care out of pocket for basic health services. Chronic diseases and complex hospital interventions are not covered. 148. This year the pricing and subsidisation approach has changed, in order to promote a more equitable approach and improve the financial sustainability of the scheme. The premium amount to be paid will be based on a household’s Ubudehe category. Those in Ubudehe category 1 and 2 will have their fee subsidised by the government (2,000 will be paid to the mutuelle on their behalf); households in Category 3 will pay 3,000 per member, and households in Ubudehe categories 4 through 6 will pay 7,000 per member. 149. This community-based health insurance is therefore one of the two pillars of the overall Health Financing Policy of 2009; while the mutuelle addresses demand-side barriers to affordable access to health care, the second pillar is to improve the supply side through increased efficiency in the use of resources and coverage of high-impact interventions. Therefore, although for purposes of the social protection strategy it is helpful to consider the mutuelle separately since it plays an explicit role in mitigating household risks around health expenditure, it is important to remember that it is just one part of the overall financing of the health system as a whole by government and donors. Since the mutuelle is not actuarially sound - in that the contributions are set at a level much lower than the actual costs – the mutuelle should be seen more as one of the ways in which government subsidies health care (along with performance-based approaches to financing of health centres, expanding geographical access through the 74 construction of new facilities, provision of drugs and consumables, etc.), rather than a stand-alone programme37. 4.3 Social development programmes 4.3.1 VUP Financial Services Programme description 150. The third component of VUP – Financial Services - is not part of the ‘core’ social protection sector, but it is nevertheless considered to be a complementary programme. Within this, there are 3 sub-components, namely microcredit (known as the Ubudehe Credit Scheme, or UCS), the Challenge Fund (a set of matching grants for purchase and/or construction of assets requiring large capital inputs), and financial literacy training. It was launched after the two other VUP components, with the UCS starting in January 2010. 151. The target population for the financial services is wider than that of the Direct Support and Public Works components, since it reaches households in all Ubudehe categories. Households in categories 1, 2, and 3 can apply for individual, group, or cooperative loans, while even households in categories 4, 5, and 6 can apply for group or cooperative loans as long as households from categories 1, 2, or 3 are also part of the groups/cooperatives. In addition to meeting these criteria in terms of Ubudehe categories, the individuals/groups taking out the loans must submit a proposal for a viable income generating activity – in other words, being able to demonstrate that the loan will be repaid from the activities – and show that they have the capacity to manage the loan and the activities. The borrower then signs a performance contract with either the village, cell, or sector as appropriate. 37 See Kalk et al (2005) for a good discussion. 75 152. Maximum loan amounts depend on the size of the group, ranging from RwF60,000 for an individual to RwF100,000 per person for a group of 11 or more. Loans must be repaid within 12 months at an interest rate of 2% per year. Given the objectives of reaching poor households, there are no guarantees or collateral required. Instead, mutual trust and social influences are expected to ensure that loans are repaid. Progress in implementation 153. By the end of the 2009/2010 Fiscal year (i.e. after the first 6 months of implementation), there were already a total of 55,675 individual beneficiaries through 14,536 loans. The average loan amount per beneficiary was RwF77,480, which is slightly more than the average amount of transfers paid in wages to beneficiaries of the VUP public works component. Challenges so far and lessons learned 154. Although the Financial Services component is quite new, there are some initial lessons that have already been learned relating to some early challenges. These included the need to carefully consider:  The timing of the disbursements of the loans, since the timing of the purchase of inputs is often time-sensitive if the income-generating activities are to be successful, particularly those loans relating to agricultural production.  How to handle cases of default; given that the Credit Scheme is intended to be a poverty reduction programme first and foremost, thought must be given as to how to handle those cases where the income generating activity is not successful and borrowers are unable to repay.  Increasing repayment rates, which in the first round of loans was quite low. 76 4.3.2 FARG income generating and shelter programmes 155. In addition to the provision of Direct Support transfers as described above, FARG also has other programme components that fall under the ‘complementary’ heading, including support for shelter and income generating activities. 156. For the shelter component, houses were constructed, rehabilitated, and in limited cases purchased, for needy genocide survivors. From 1998 to 2010, over 14,857 houses were constructed and a further 1,396 rehabilitated. 157. For income generating activities, grants are provided to groups of genocide survivors for various projects. These grants have not been steady over time, with 8,595 beneficiaries from 1998-2000 and then a gap until 2005-2008, when there were a few hundred beneficiaries per year. In 2010 there was another spike in the number of beneficiaries, reaching 2,000 in that year.38 4.3.3 MINAGRI One Cow programme 158. In May 2006, the Government launched the ‘One Cow per Poor Family’ programme, also known as Girinka, and it became operational in November 2006. The objectives are to reduce malnutrition particularly among children and increase the incomes of poor framers by providing them with a cow whose milk they can both consume and sell. 159. Households are provided with a cow (a cross breed that produces more milk than traditional domestic breeds) as well as support for primary animal care including artificial insemination, seeds for animal feed, and some veterinary treatment and vaccinations for the animals. 38 For validation – need to verify this spike in beneficiary numbers, since the amount of expenditure does not increase proportionately (indeed, it is lower in 2010 with 2,000 beneficiaries than in 2009 with only 220 beneficiaries). Perhaps a typo in the 2010 data? 77 160. The programme is supported by partners who support delivery in different ways:  MINAGRI has distributed over 30,000 cows through the Rwanda Animal Resources Development Authority (RARDA) “Projet Appui au Développement d’Elevage Bovin Latier (PADEBL) and support to the PSTA Project .  Donor and local businesspeople have directly distributed cows through the District Community Development Programme, NGOs and local banks.  District governments have also bought and distributed cows through their own resources. 161. Given the many different delivery mechanisms, Girinka has therefore developed more as a model of providing households with cows rather than a unified programme. Households being provided with cows through the MINAGRI programmes are selected based on criteria of being needy but also of having at least 0.75 hectare of land and being able to support the animal (as well as not having already received an animal). Across all the different delivery options listed above, the objective is to address poverty and therefore to target households that are most vulnerable. 162. Once the cow has given birth to the first calf, the household then provides the calf back to the programme so that it can be passed onto another needy household. As such, the programme ensures that the benefits are continually paid forward, with a total of over 115,000 families having benefited39. 4.3.4 Ubudehe40 163. The Ubudehe programme began in earnest in 2005, with an objective of reducing poverty in the south-east of the country, as part of the Decentralised Programme for 39 Source: Programme data. 40 The source for this section is the Ubudehe evaluation. 78 Rural Poverty Reduction (DPRPR). The first phase ran from 2005-2006, and was scaled up for a second phase in 2007-8 to reach over 14,000 villages across the country. 164. The Ubudehe component involved a participatory poverty assessment, where villagers articulate their understanding of poverty through a social mapping exercise. Through this process, they elaborate their own priorities for development and then create a Community Development Plan. The Ubudehe programme provides funding for the priority activity in the plan (for example, reconstructing a school) and then the community undertake the work to complete it. 165. In addition, the community identifies one household per village – deemed the poorest and most vulnerable – to receive a grant of RwF60,000 (roughly the cost of the minimum food basket for one adult per year at the time). The idea is that the recipient will use the grant for some kind of income-generating activity, and then will be able to ‘pay forward’ a part of the return to another needy household. The most common investment for beneficiary households was to buy livestock (frequently a goat), and then the first kid born would be given to another household in the village, and so on. 166. Although a formal impact assessment was not undertaken, a qualitative evaluation found that the programme had a positive impact on beneficiary households. While nearly all of those surveyed said that before the programme they had little or no capacity to nourish their family, after the programme, only 10% had little capacity, while the remainder said they were able to nourish their families “fairly” or “well” – a marked improvement. Similar results were found for the ability to buy clothes and to pay school fees and health expenses. 97% of respondents said they were less poor as a result of the programme. 167. Execution of the programme, funded by the Government and the EU, continued through 2009 and was then brought to a close. The Ubudehe approach to participatory targeting has continued, and indeed grown, through the VUP programme, although the 79 community development activity and support to individual households stopped. These will resume again in 2012/2013. 4.4 Complementary services 4.4.1 MIGEPROF support for OVCs 168. MIGEPROF is responsible for a range of activities related to child development, child protection, and gender. Much of the support for children involves a particular emphasis on Orphans and Vulnerable Children (OVCs). MIGEPROF’s activities for OVCs include41:  Skills development and support for income generating activities: in 2011, 500 OVCs received support to attend vocational training and received a start-up kit so they could establish their own small business, and 10 projects were funded through the Association des Orphelines Chef de Ménages.  Early Child Development (ECD): MIGEPROF supported around 50 ECD centres in 10 districts with materials and management training.  Orphanages: 27 orphanages were provided with financial support. 4.5 Achievements and gaps 169. What is perhaps most striking from this discussion is the scale of the achievement in the social protection sector, which has included on one hand the development, implementation, and scale-up of a fully nationally-owned and significantly government-funded transfer programme in VUP, and on the other hand the development and cabinet approval of a social protection strategy and implementation 41 MIGEPROF’s activities are gleaned from the Office of the Prime Minister (2011) Government Annual Report 2010-2011. 80 plan. In just five years, Rwanda has therefore managed to achieve something incredibly rare: political commitment, the allocation of resources, delivery of benefits, and, perhaps even more impressive, a clear guiding strategy setting out the future growth path of the sector. 170. The gaps in the sector must therefore be seen in this context: the social protection sector is growing progressively as envisaged in the social protection strategy, and gaps are related mainly to the rate of scale-up and development of new programmes and to budgetary constraints. The main gaps are of two types:  Geographical, in the sense that VUP is now operating in 120 out of the 416 sectors out of the country.  Coverage in terms of poverty levels, since VUP only targets households in Ubudehe categories 1 and 2, reaching only a portion of the extremely poor households in the country. This is, however, due to budgetary constraints, and as a result the targeting process aims to identify those in the most need. 171. While budgetary constraints will always remain in some form, the National Social Protection Strategy envisages a progressive realisation of the social protection floor over the next ten years, as indicated in Chapter 3 above. As such, the strategy and related implementation plan have already addressed the main gaps in coverage over time. 172. Remaining gaps related to design and implementation problems around targeting and disbursements and systems issues related to monitoring, evaluation, public finance management identified in Chapter 3 will be addressed in more detail in the conclusion, while some of the bigger questions relating to institutional arrangements – both within government and in government/donor coordination – will be addressed in the following discussion in Chapter 5. 81 173. Aside from these gaps covered elsewhere, one of the most important ‘works in progress’ that stretches across the whole social protection sector is the development of an integrated Management Information System (MIS). The need for a more coordinated and consistent MIS – to fill not only specific gaps in each of the programmes but also contribute to the overall needs of MINALOC to supervise performance of the sector – was strongly recognised in the Strategy and Implementation. Currently expertise has been sought to map out the best technical options to achieve the desired integration. Ultimately the precise technical specification (for example, whether a single ‘master’ database versus coordinated programme-specific databases that ‘talk’ to each other) should depend on the overall institutional structure discussed further in the following chapter. 4.6 Overlaps and duplications 174. Given the clear strategic direction for the sector already existing, it is actually the current overlaps and duplications that require more urgent attention than the gaps. These overlaps are most obvious in the areas of direct support transfers (from VUP, FARG, and RDRC), housing support, and income-generating activities. 175. Consolidation options for direct support are discussed in detail in the following chapter, as this is a core part of social protection and deserves a full analysis. Housing and income-generating activities are, by contrast, complementary activities that should therefore be the main responsibility of other sectors rather than a core focus of the social protection sector. In practice, however, these do not fall squarely inside of any one sector and therefore tend to continue to be monitored mainly under social protection. 176. Support for housing comes from the following range of programmes:  FARG shelter component.  MINALOC decentralised social protection spending (until current year). 82  MINALOC housing and urban settlements.  Refugees and returnee programmes; and  Housing for demobilised disabled former combatants. 177. Income-generating activities also include a range of different programme and spending lines:  FARG income generation components.  MINALOC decentralised social protection spending (starting in 2011/2012 fiscal year with new guidelines).  Programmes focused on youth.  Programmes focused on women (largely donor funded).  RDRC programmes for former combatants. 178. Each of these programmes is implemented entirely separately, with separate uncoordinated funding streams, different target populations, and different mechanisms for selecting beneficiaries, as illustrated below. Table 12: Summary of shelter and income-generating activities Main Type of Targeting guidelines / funding Programme benefit Target population mechanism source Needy genocide FARG Shelter Shelter survivors Community selection GoR FARG Income Generating Needy genocide Activities IGA survivors Community selection GoR MINALOC social protection None - largely ad hoc sub-programmes (ended in requests from districts 2010/2011) Shelter/IGA Poor/vulnerable to MINALOC GoR MINALOC Decentralised line Ubudehe categories 1 item IGA Poor/vulnerable and 2 GoR District own revenues Shelter Poor/vulnerable None - districts decide GoR MINALOC housing and urban settlements Shelter Unknown unknown GoR IGAs for women IGA Women unknown Donors IGAs for youth IGA Youth unknown Donors Shelter for Refugees Shelter Refugees unknown Donors 83 179. The harmonisation of the ‘core’ social assistance programmes as discussed in more detail in the following chapter should of course be the biggest priority for the sector. However, the proliferation of these activities in shelter and income generation suggest that at least some initial informal coordination should be orchestrated through the social protection working group, perhaps through a sub-committee established for this purpose. 84 5. Towards the next phase of the social protection sector: hurdles to overcome and a vision for the future 180. As the previous two chapters illustrate, the social protection sector in Rwanda has been rapidly maturing, from a fragmented, weakly defined and generally uncoordinated sector five years ago to a more strategic, government-led, and coordinated one of the present. With this progress, however, comes a new set of challenges as the sector moves forward and continues to evolve. These include:  Harmonisation and consolidation of transfer programmes.  Decentralisation and implications for social protection.  Institutional arrangements for policy leadership and oversight; and  Donor coordination arrangements within a SWAP. 181. In principle there is a general consensus on the fact that these hurdles will need to be overcome in pursuit of the new strategy as envisaged in the Implementation Plan. However, in practice ‘the devil is in the detail’. This chapter will address each hurdle in turn, outlining the key issues that will need to be considered, documenting available evidence, and presenting a set of options for MINALOC and development partner consideration. 5.1 Harmonisation and streamlining of programmes 182. The Strategy, Implementation Plan, and indeed the general discussions within the social protection working group all point towards the need to harmonise and ‘streamline’ the existing programmes in order to achieve a more coherent, coordinated, and consistent system. Previous analytical work in the sector identified a range of areas of overlap and duplication that would seem to be ripe for such consolidation, including FARG Direct Support and support for disabled demobilised soldiers. 85 183. Of these two, the highest priority of these is the integration of FARG and VUP Direct Support programmes, which are seen as being problematic because they are – at least on the surface – similar in terms of providing unconditional cash transfers but are inconsistent in their targeting approach, transfer amounts, and payment mechanisms. At its simplest, consolidation of these two programmes was envisaged in a way that would see FARG Direct Support ending in areas where VUP is operating, with the idea that FARG DS beneficiaries would then be covered by VUP instead. For all practical purposes, this would essentially involve a simple discontinuation of FARG Direct Support where VUP is operating, with FARG DS fully phased out once VUP has scaled up to all sectors in the country. 184. This view of consolidation hinges on the assumptions that: I. FARG Direct Support beneficiaries who are genuinely needy would be identified as such by the Ubudehe targeting process used for VUP; in other words, they would be in Ubudehe categories 1 or 2; and II. Current FaRG DS beneficiaries who are not identified as Ubudehe categories 1 or 2 are essentially ‘mis-targeted’ because they are not truly needy, and therefore support to them could be discontinued. 185. Before implementing the envisaged consolidation, it is therefore important to test these assumptions and estimate the scale of the impacts on current FARG beneficiaries. If the majority of FARG DS beneficiaries are already in Ubudehe categories 1 and 2, then the consolidation could imply a fairly straightforward shift of FARG DS beneficiaries onto VUP rolls. If, on the other hand, a large percentage of current FARG DS beneficiaries would not already be eligible for VUP, then it will be important to consider the implications of the consolidation more fully and make policy choices that as much as possible answer to issues emanating from the analysis. 86 186. The first set of questions is therefore essentially empirical: what is the profile of current FARG DS beneficiaries? How many will lose out if FARG DS is dissolved, and how many of those losing out are truly needy? The answers to these empirical questions will feed into the second set of questions around government’s current commitments (statutory or otherwise) to genocide survivors and the compatibility between the objectives of programme consolidation with the objectives of support to genocide survivors. FARG Direct Support beneficiary matching exercise 187. In order to address the empirical questions, what was needed was a way to match the FARG Direct Support beneficiary list with the Ubudehe targeting list in order to analyse the distribution of FARG DS beneficiaries by Ubudehe category. As this is quite data intensive, a matching exercise was undertaken for a sample three districts (Musanze, Kicukiro, and Bugesera). A more detailed description of the methodology is provided in Annex B. 188. The analysis revealed that, on average across the three districts, only 4% of FARG Direct Support beneficiaries fell in Ubudehe category 1 and 15% in category 2. In total, therefore, only 19% of current FARG DS beneficiaries would be covered by VUP Direct Support or Public Works, while in Musanze it would be only 11%. Equally striking is the fact that a very large percentage of current FARG DS beneficiary households fall into category 3: 51% in total, and up to 57% in Bugesera. About a quarter of households fall into category 4 and only 5% fall into category 5, although in Musanze these figures are 33% and 12% respectively. 87 Figure 13: Distribution of matched FARG beneficiaries by Ubudehe category and district 189. Within districts, there is also some variation across sectors in the distribution, as illustrated in the figure below for Kicukiro. In Nyarugunga and Gahanga over 30% of beneficiaries are in categories 1 and 2, whereas in Niboye, Masaka, and Kicukiro it is less than 10%. Similar disparities across sectors are found in the other two districts. 88 Figure 14: Distribution of matched FARG beneficiaries by Ubudehe category in selected sectors in Kicukiro district 190. Of course it is important to be careful in the interpretation of the data, given that only about 40% of beneficiaries have been matched successfully. If there were any systematic bias in the matching of beneficiaries then there may be a concern that these averages based on the 40% sample of FARG beneficiaries were not representative of the overall FARG DS beneficiary population. This would apply mainly to child beneficiaries, since they are the ones without ID numbers, as well as adults whose ID number was entered incorrectly, who also had other individuals in the same cell with the same name (and were therefore not able to be uniquely matched). It would therefore not seem, a priori, that there would be any major concerns about systematic bias in the matched sample, since whether or not individuals had errors in their IDs or common names would not seem to be correlated with Ubudehe category. 89 191. While bearing these caveats in mind, it is clear that ‘consolidation’ of FARG DS into VUP would essentially mean that potentially 80% of current beneficiaries would lose out entirely in the transition. This then prompts some further questions: are they mis-targeted by FARG? Are these beneficiaries in category 3-6 truly ‘needy’? FARG targeting: consistent with objectives? 192. To answer the first question about proper targeting of FARG DS, we can return to the eligibility criteria, remembering that there are essentially three categories of beneficiaries: orphans, disabled adults, and the elderly who are without means of support. Unfortunately, we do not have detailed enough data on beneficiary characteristics to establish which criteria a given beneficiary meets. However, an analysis of the age profile of beneficiaries found that many of the current beneficiaries are likely to be children who would ‘graduate’ out of the programme once they turn 2142. The remainder of adult beneficiaries would then (according to the eligibility criteria) be individuals living with disabilities and the elderly. 193. Based on available data, therefore, there is no immediately apparent evidence of obvious mis-targeting. However, further monitoring and evaluation efforts should be placed on better understanding the profile of FARG beneficiaries and their eligibility. Poverty of FARG beneficiaries 194. In terms of the assessment of whether FARG DS beneficiaries are truly ‘needy’, it might be tempting to assume that those households in Ubudehe category 3 or higher are not poor and could therefore be dropped from support. It is important to remember, however, that the Ubudehe categories do not correspond perfectly with consumption poverty. The results of the VUP Survey showed that 52% of those in Ubudehe categories 42 Although it is not currently entirely clear why there are currently some younger children (born long after 1994) seem to be included as per the matching exercise. Is this because some younger orphans (not orphaned immediately as a result of the genocide but perhaps whose parents where genocide survivors who then passed away later due to other causes)? Or are these errors in the matching data? Or incorrectly targeted beneficiaries? 90 1 & 2 were extremely poor in terms of consumption (meaning that, conversely, 48% of these two categories were not extremely poor in terms of consumption), while 39% of those in category 3 were also extremely poor. Furthermore, since category 3 is much larger as a percent of the population, fully 54% of the extreme poor were actually found in category 3, and only 34% in Categories 1 & 2. Even in categories 4-6 there were households who were extremely poor in terms of consumption, with 12% of the total extremely poor found in these higher Ubudehe categories. 195. Given these incongruencies between Ubudehe measures of poverty and measures of consumption poverty, it would be wrong to conclude that simply because FARG DS beneficiaries are found in Ubudehe categories 3-6 they are not legitimately needy. Shifting exclusively to the VUP eligibility criteria would mean that many legitimate recipients of FARG Direct Support would therefore be likely to lose out as a result of imperfections in the Ubudehe classification. In other words, the FARG targeting process may be correctly identifying some truly extremely poor households that are incorrectly identified by the Ubudehe category. 196. Furthermore, those who are correctly identified as Ubudehe category 3 are currently not included under VUP because of budgetary limitations, while they may be included under FARG Direct Support. Again, this does not mean that the FARG beneficiaries in Ubudehe category 3 are not needy, merely that the threshold for inclusion may be slightly higher than what the VUP budgets can currently handle. 197. Either way, there is in practice often very little to distinguish between those in Ubudehe category 2 and 3, and cutting off benefits for those in Category 3 unequivocally is likely to jeopardise the inclusion of many legitimately needy current beneficiaries. Other considerations 198. Beyond what we can glean from the quantitative analysis, there are some other important considerations: 91  Statutory commitments: The FARG law is somewhat vague, however at present it is worded such that the neediest should be selected first, not that only the neediest are the only ones entitled to assistance. It will be important to ensure that any proposed programme changes are either consistent with the current legislative wording or that appropriate changes to the statutory commitments would be articulated.  Equity: Phasing the consolidation of FARG into VUP as new sectors are incorporated into VUP would lead to large inequities across sectors. This is particularly important given the scale of current FARG DS beneficiaries who would be excluded from VUP.  Rationale for individual transfers under FARG: assistance supporting individual orphans, for example, which is relevant even if the rest of the household does not receive additional transfers.  Related to this, work conditionality for those households that would only be eligible for VUP Public Works (not VUP Direct Support) due to the presence of an able-bodied member. This could be, for example, a household with an elderly person currently receiving FARG Direct Support (as an individual) but who is living with someone of working age. If the working household member is unable (due to other commitments, for example) or unwilling to take up public works employment, then again it is de facto the discontinuation of the benefit to the original beneficiary. The same goes for households who receive FARG DS because they have taken in an orphan of the genocide: shifting them onto VUP Public Works would change the nature of the benefit entirely, making it no longer a grant to help offset the additional costs of the orphan. Also, public works employment is not guaranteed to every eligible household every year: due to financing limitations and limitations on the number of projects which can be implemented, households may only be granted employment one out of every three years. 92 Implications and potential options 199. The preceding analysis has shown that there are no simple answers: a simple integration of FARG Direct Support into VUP would exclude a large number of current beneficiaries, many of whom are likely to be legitimately needy. Ultimately the targeting approach of VUP is not consistent enough with that of FARG, and the programmes currently serve different populations with slightly different objectives (for example individual versus household coverage). In this case, ‘consolidation’ will actually mean discontinuing support. Unless policy commitments are actively changed (for example, there is a clear policy decision that government is no longer obligated to support genocide survivors who are orphans, widows, or disabled), then it is not possible at the current time to simply fold them into VUP. 200. However, this is not to say that there are no other potential options. Given the demographic breakdown of FARG beneficiaries, with a portion of them being orphans whose eligibility will end by 2015 and the remainder being adults with disabilities and elderly, there may be scope to re-think the consolidation. Rather than forcing integration of two programmes with very different target groups and beneficiary selection processes in the short term, another option would be to consider consolidation in the medium term instead. The implementation plan has a feasibility study for an old-age grant as a major activity for this year, and this would be an ideal time to assess the feasibility of consolidating FARG Direct Support with an old age/disability grant in future. The consolidation of these two programmes would be much more seamless than integrating FARG and VUP at present. 201. In the short term, integration of FARG DS and VUP could still be partially achieved through the consolidation of delivery mechanisms. So, in other words, for the next couple financial years:  FARG DS and VUP targeting would still continue as they are now as separate processes. 93  In sectors where VUP is operating, FARG DS lists would be provided to VUP.  VUP would be responsible for the payment of beneficiaries, financial reporting, and monitoring. Figure 15: Ease of consolidation of FARG Direct Support with VUP and Old Age Grant 5.2 Decentralisation and implications for social protection 202. Decentralisation is at the very core of governance; at its heart are issues of rights (who is entitled to what), responsibilities (who must deliver what), and accountabilities (who is held accountable, by whom, and how). Within these governance relationships, the nature of decentralisation also largely determines the efficiency and effectiveness of spending, defining how much is spent, who has discretion over inputs, outputs, and outcomes, and what incentives are in place for delivery. 203. The answers to these questions are particularly important in the social sectors, where services are delivered at local level but there is a large concern with ensuring equity across the country as well as efficiency. Rwanda has been implementing a decentralisation strategy over the last ten years, and has now entered the third phase with the new Fiscal Decentralisation Strategy (2011-2015) The first phase focused on political decentralisation and setting out the broad assignment of functions across levels of government, while the second phase focused on streamlining local structures 94 (reducing the number of districts and provinces), redefining responsibilities, and building capacity for increased fiscal decentralisation. The focus now in the third phase is on consolidating the reforms undertaken so far, improving implementation of the policy rather than setting new objectives. 204. One of the main areas of emphasis in the new strategy is on changing the balance of control over expenditure decisions at local levels. At the moment, the majority of decentralised spending is tightly earmarked, constraining the discretion of districts, while at the same time there are concerns that the current assignment of responsibilities and burden of administrative procedures tend to overload local government staff in the context of human resource constraints. 205. The decentralisation strategy therefore articulates an objective of reducing the extent of earmarking (or tying) of funds and allowing more discretion at district level. However, sectoral ministries are still responsible for the “establishment of sectoral policies, strategies and guidelines for local service delivery as well as the provision of fiscal transfers to enable services delivery, and monitoring of budget execution and performance in their respective sectors”. The sectoral ministries also play a significant role in the provision of specific technical assistance and capacity building to the local governments relevant to their sectoral expertise. 206. A recent review of decentralisation (MINALOC 2010: x) also points out that sectoral decentralisation is ‘largely unsatisfactory, with only Education and Health who have decentralised a measure of public service delivery to the local government. The process of sectoral decentralisation will have to be accelerated in phase three of implementing decentralisation policy.” (italics in original) 207. This obviously raises some important questions for social protection, and indeed all of the social sectors. Social protection is normally not a function that is fully decentralised, and getting the balance right between central and local mandates, funding flows, and accountability mechanisms is perhaps more challenging than in some 95 other sectors, since social protection places an explicit focus on issues of equity through the definition of entitlements to particular transfers and benefits. 208. Taking this third phase of decentralisation forward in Rwanda – and ensuring the optimal results in terms of equity and efficiency – will require some further nuanced consideration rather than a simplistic approach to decentralisation. It is important to note that - contrary to what is often presumed in the literature on decentralisation - when it comes to social policy there is in fact a large degree of central control over spending even in developed countries: ‘more’ decentralisation is not always better. It is therefore worthwhile to draw some lessons from other countries that may be relevant for consideration for Rwanda as MINECOFIN and MINALOC begin to put the latest decentralisation strategy into practice. What do we mean by decentralisation? 209. Before drawing these lessons, however, it is helpful to start with some basic definitions and the underlying rationale for decentralisation43. For the purposes of this assessment it is important to focus on the distinctions between the transfer of responsibilities, the level of discretion over decisions at the local level, and the implications for accountability. We can separate some of these decisions into three main dimensions (political, administrative, and fiscal), and for each of these there is a spectrum between more and less centralised systems. 210. The first dimension is related to political decentralisation:  Deconcentration, defined as the transfer of responsibility from central ministries to field offices or autonomous agencies, so that implementation is closer to 43 Unfortunately, there are no standard definitions for how to measure the spectrum of arrangements between more and less centralised systems, and concepts such as decentralisation, deconcentration, devolution, and delegation may have different meanings in different countries and across different authors. See UNDP (1999) for a review of the literature and to get a sense of the very wide divergence in definitions used across authors. 96 citizens but nevertheless remains part of central government (in other words, local officials are not elected); and  Decentralisation, defined as the transfer of responsibility to democratically independent (elected) lower levels of government, which provides some level of managerial discretion but not necessarily financial independence. 211. Rwanda clearly lies on the fully ‘decentralised’ end of the spectrum with respect to the political dimension. 212. The second dimension is administrative: whether local officials have discretion over managerial inputs and processes. This often refers largely to the ability to hire and fire staff, decide what staffing structure is put in place, and what internal processes are used. On this dimension Rwanda sits towards the ‘local control’ end of the spectrum, with district councils able to make decisions over hiring and firing in most posts44. 213. The final dimension is fiscal, whether local officials have the power to raise revenue and decide on expenditures. This depends both on the source of funding (whether from central tax resources or local ‘own revenue’) and the extent of central control over spending decisions (for example, through setting mandatory/statutory functions, minimum standards, etc.). See the Box below for a description of different types of funding and the implications for local control. 44 Provinces have centrally appointed governors, however provinces have extremely limited functions. 97 Box 4: Understanding fiscal decentralisation Local governments have four main sources of revenue: ‘own-source’ revenue from taxes collected by local governments directly; transfers from central government, local-government borrowing, and payments from users for services. There is a range of different types of grants, which imply different levels of autonomy for local levels as well as different types of accountability.  Specific grants: Also referred to as ‘earmarked’ grants, these are related to mandatory tasks, and can only be used for the purpose specified (which can vary in terms of how precisely this is defined).  General grants: Also referred to as ‘block grants’, these can be used for any purpose as decided by the local government. Earmarked grants are used where it is desirable to ensure minimum standards of delivery, for both quality and equity reasons, since these provide the central level with a degree of control over outcomes. General grants, on the other hand, may be preferred by local governments because they allow more flexibility, but they also limit the ability for local governments to be held accountable for results by central government. 214. On the first two dimensions (political and administrative) Rwanda sits towards the ‘local control’ end of the spectrum. With respect to fiscal decentralisation, however, the majority of spending is still fairly tightly controlled through earmarked transfers. Districts therefore have little discretion over the purpose of much of their expenditure or even how it should be spent (for example on salaries versus functional costs). Putting it all together: the 3Fs – funds, functions, and functionaries 215. Taken simplistically, political, administrative, and fiscal decentralisation should in theory be consistent. This is referred in some places as the 3Fs of decentralisation: funds must match the functions that have been assigned to sub-national governments, and there must be the requisite control over the functionaries in place to carry out responsibilities. 216. In practice, however, in most cases these three dimensions do not necessarily move in tandem nor will they be the same for all functions even within a given country, providing a wide range of experiences even amongst contexts referred to as 98 ‘decentralised’45. Furthermore, decentralisation and deconcentration are very likely to exist at the same time; some functions will be decentralised while others will be deconcentrated to semi-autonomous agencies or local-level offices of central line ministries. 217. It is therefore important to continually bear in mind three sets of questions relating to the 3Fs:  Division of responsibilities: Which level of government (or agency) has responsibility for a given functional area? How are these responsibilities shared?  Financing: From which sources are the responsibilities funded (locally-collected taxes, user fees, taxes collected by the centre but shared with local level, central grants)? How much discretion does the decentralised agency have over spending decisions?  Administration: Who has control over decisions to hire, fire, and promote staff, set salary levels and working conditions? 218. In order to help determine the best answer to these questions for an individual country, it is possible to distinguish between three types of responsibility:  Those that belong wholly to the centre, such as diplomacy, defence, and monetary policy.  Those that are the exclusive domain of the local level, such as town planning, local transport, sewage systems and waste collection, etc. often referred to as ‘classic public services’. 45 For example, even in the United States where education is the responsibility of states and counties (and therefore in theory not an area for federal interference), the nature of education financing (where a large portion of funds come from the federal level) provides a strong de facto role for the centre in education policy, through a set of conditions on the access to and use of the funds. 99  Those that are shared tasks and responsibilities, either between different tiers of central government (i.e. deconcentrated) or between central and sub-national government (decentralised). In most developed countries, implementation of social policies (health, education, and social assistance) falls under this category. 219. It is interesting to note that the large growth in local government spending as a share of the total across developed countries that has occurred in recent years has been largely due to these latter shared responsibilities, with local governments playing an increasingly important role as the ‘handmaiden of the welfare state’. The distribution of responsibilities in most countries is often quite complex, and the lines between central and local responsibilities are often blurred. The sharing of functions might be explicit (through a division of responsibilities specified in the constitution or legislation) or implicit (where central government funds activities in a manner that gives them control over decisions in implementation). 220. While Rwanda certainly is not a ‘welfare state’ in the same way as most of the OECD, it has a very strong development agenda, and the focus on human capital development and social service delivery means that in practice the role of local governments in these areas is not dissimilar to developed countries. 221. Central government involvement in social policy areas, while contrary to the simplistic typology of ‘good’ decentralization, is therefore not surprising given the importance of social services to the voting public. Ongoing debate over the best balance of autonomy and accountability to the centre is therefore likely to continue – as well it should - since there is no easy or ‘correct’ solution to these issues. Simple calls for further sectoral ‘devolution’ are unlikely to be fully appropriate for many of the social sectors, including in particular social protection. A sector-specific approach to the degree of decentralisation need not be inconsistent with Rwanda’s stated goals on decentralisation. To best understand the optimal level of decentralisation, it is important to remember the rationale for decentralisation and how it applies to social protection. 100 Why decentralise at all, and why decentralise social protection? 222. The rationale for decentralisation in general, particularly for developing countries, is generally associated with increasing accountability to citizens by moving government ‘closer’ to the people and increasing the efficiency of service delivery. Efficien cy is expected to increase both in allocative terms – by reflecting the needs and preferences of the poor through their greater participation – and in operational terms – by allowing local government greater control over how they implement their responsibilities and by allowing citizens to hold them to account more easily than they could central government. The expected outcomes are therefore greater access to more efficient services (lower corruption, more innovation, better poverty targeting and higher ‘v alue for money’), improved ‘voice’ and reduced vulnerability of the poor46. According to this logic, services should be decentralised to the lowest unit of government that can efficiently deliver (otherwise known as the principle of subsidiarity47), with central government retaining a role for setting standards in order to ensure equity, and local governments should be given as much discretion over their finances as possible. 223. Within this framework, decentralisation is thought to increase efficiency of service delivery through the following channels:  As local elected officials are considered to be ‘closer’ to the people, it is expected that voice is strengthened through decentralisation.  Decentralisation increases ‘choice’ through ‘exit’, whereby citizens are able to literally ‘vote with their feet’ and move to other areas where services are seen to be better. This increases competition amongst local governments to perform 46 See for example Jütting et al (2004: 10) and World Bank (2004: 186-190). 47 More formally, subsidiarity involves placing service delivery responsibility at the smallest unit where marginal costs equal marginal benefits. This more thorough definition includes the idea that externalities (where costs or benefits are incurred by someone else) are limited in order to ensure efficient service delivery. Immunisations and other public health interventions, for example, are generally not well suited for full decentralisation, since local governments would incur costs, but other areas would benefit. This would lead to sub-optimal levels of provision. 101 better by strengthening the ‘compact’ and management. Efficiency is therefore driven by innovation and experimentation at the local level. 224. However, two conditions are required for these efficiency gains to be realised. First, citizens must have access to relevant information on performance across local governments (or providers, if decentralised directly to them) and they must be able to move and exercise their own choice. In practice, in many places this will be difficult, since the poor in particular are not able to migrate, and in rural areas there may not be enough providers to have any real choice. Second, local governments and providers must be allowed the freedom to innovate and experiment; otherwise the ‘choice’ between them will be limited. 225. On the other hand, the more choice and discretion there is at local level, the greater the inequities will be across local governments: some districts could perform better, but some could perform worse. Without some sort of central control, there is no way to guarantee delivery of services of a particular quality standard. As a result, the further functions are fully ‘decentralised’ the less they become an ‘entitlement’ guaranteed by central government. 226. When it comes to social protection, there are very few areas over which local governments should have full control. This is because:  Entitlements are determined nationally, so that citizens in each and every district should normally have the same entitlements granted to them on equity grounds. Ceding local control over the design of transfer programmes is therefore not advised.  Transfers should be equalised nationally – in other words, funded from central revenue not local revenue, again to ensure equity across the country. 102 227. This rationale for full ‘decentralisation’ (meaning simply granting greater local autonomy or ‘devolution’48), as appealing as it may seem on the surface, glosses over the many complexities of decentralisation as well as the implications for rights, governance, and accountability that are implied in the simple ‘more is better’ approach. As Rwanda now moves forward with its third phase of decentralisation, with a particular emphasis on deciding the optimal distribution of functional responsibilities, appropriate fiscal transfer arrangements, and mechanisms of accountability, it is the perfect time for the social protection sector to review these issues and enter into discussions with MINECOFIN, Cabinet, and Parliament on the arrangements that will be most appropriate for social protection both now and in the medium term. Decentralization in Rwanda’s social protection sector 228. Currently the formal assignment of responsibilities for social protection does not clearly specify which functions are shared or the mechanisms for ensuring joint accountability. This is partly a function of the multi-layered ‘decentralisation’ that exists in the current system, with some responsibilities devolved to agencies (such as FARG and now RLDSF) which then enlist the districts for some aspects of delivery, and some responsibilities granted to districts directly (through earmarked transfers). 229. In the current (2011-2012) fiscal year, changes have been made in the flows of funds to districts, with many transfers now going directly from MINECOFIN to districts under earmarked line items, rather than passing through, for example, RLDSF as an agency. 48 Often the discussions of decentralisation in Rwanda refer to ‘devolution’ of functional areas. It is important to note that ‘devolution’ generally refers to the full transfer of responsibility to lower levels of government, giving them full autonomy over policy as well as expenditure, allowing them to determine what is provided within that functional area, how much is spent, and how it is spent. It is however unlikely that this is really what is intended in most instances. 103  FARG: Responsibilities for delivery are deconcentrated in an agency model to FARG. Districts (through the Vice Mayor in charge of Social Affairs) have some responsibilities for implementation, in terms of facilitating the beneficiary selection and identification process, but FARG disburses the majority of the funds directly. In the current fiscal year, additional personnel has been placed in districts to ensure proper targeting and certification of educational institutions receiving funding for children sponsored by FARG, however it is not clear to what extent these are district staff or merely FARG staff who are placed at the local level.  VUP: As with FARG, the primary deconcentration of responsibilities is from MINALOC to RLDSF for the delivery of the VUP programme. In the case of VUP, however, there are two staff members placed in each sector in which VUP operates, whose salaries were, up to the current year, paid out of the VUP budget. Funds were transferred from VUP to the sectors for implementation, and sectors reported back to VUP. In the new context of decentralisation, VUP allocations will now flow directly from MINECOFIN to districts, making the lines of accountability back up to RLDSF somewhat blurred and therefore somewhat diminishing the role of RLDSF as a classic ‘agency’ model.  Earmarked MINALOC transfers to districts. These have always flowed directly to districts, with the main change this fiscal year being that there are a set of guidelines that direct how the funds should be spent. Previously there has been no reporting on these funds back to MINALOC (indeed, districts have not really kept track of how these funds are spent at all). o The new guidelines now clarify the target groups (to be selected based on Ubudehe category) and the types of activities.  District own revenue: during the qualitative discussions with districts, we found that some districts are undertaking activities nominally falling under social 104 protection, although these tend to be providing housing to vulnerable groups and the provision of livestock (in efforts similar to the One Cow programme). Some outstanding issues to resolve 230. The changes to decentralization this year should in many ways improve the efficiency and effectiveness of spending, by, firstly, clarifying the purpose of the earmarked MINALOC transfers, and, secondly, reducing delays in implementation of VUP caused by funding flows needing to be routed through VUP HQ as a ‘middle -man’. However, there are some remaining issues that will need to be resolved. 231. In terms of the MINALOC decentralized transfers, the current guidelines raise some questions about how consistent this funding is with the strategy, for example: o Income generating activities are one of main types of activities to be funded: is this consistent with definition and orientation of sector in Strategy, where income generating activities (IGAs) are understood as supporting wider social development rather than ‘core’ social protection? o Furthermore, how does this funding for income-generating activities differ from objectives of VUP financial services, aside from the fact that the IGAs are grants while VUP financial services are loans? 232. On the other hand, the guidelines also raise some questions in the other direction, related to the extent to which the envisaged consolidation and harmonization is appropriate. Where there is a total harmonization of targeting approaches, combined with similar benefits, there is then a question as to whether these small funds for earmarked transfers should be simply integrated into VUP. Conversely, if there are indeed separate objectives underlying the decentralised MINALOC transfers then the guidelines should be revised for the next financial year to ensure greater clarity on the use of funds and the ultimate objectives. 105 233. Finally, if Ubudehe categories are to be used, the guidelines currently lack precision in terms of who among Ubudehe category 1 and 2 should be selected. Given the very limited resources provided through the earmarked transfers, only a small number of households can benefit, so there should be some additional criteria specified as to how districts should prioritise eligible households. 234. More generally, the guidelines and recent changes in the flows of funds to districts raise some more strategic questions related to the optimal assignment of functions across levels of government for social protection. The broad idea seems to be that social protection, as with all sectors, will be ‘decentralised’, giving more control to districts. However, as noted above, there are good reasons why social protection should be a shared function, with a strong role being retained at the centre for ensuring equity in the level and quality of provision. Returning to the conceptual framework, then, what is needed is for there to be greater clarity on:  Functions: what are the precise roles and responsibilities of MINALOC, agencies (FARG and RLDSF), and districts?  Financing flows: if funds flow directly to district, how will MINALOC and its agencies be able to ensure quality and equity in delivery? What levers will they have to intervene in under-performing areas?  Administrative control: is the current assignment of functionaries consistent with the roles and responsibilities that have been laid out? In particular, does the responsibility for decisions over hiring/firing of district-level staff working on social protection (for example, VUP or FARG) fit with the responsibility for delivery? Specifically, is FARG/RLDSF to be held accountable for delivery of these programmes? If so, are they able to do this without having administrative control over staff at the local level? 106 235. The current system – combining an agency approach with district implementation – certainly has many strengths, particularly in terms of ensuring buy-in and ownership of social protection activities at the local level and, potentially, in increasing the efficiency of spending by giving districts more direct control over disbursements. In that case, it will be helpful to simply ensure that there is greater clarity on the ‘3 Fs’ (functions, funds, and functionaries) by all stakeholders in the sector , and that these are strongly embedded in institutional arrangements. 236. However, it is also likely that the current hybrid system may need to be adjusted as the sector evolves, particularly as there is more consolidation of the systems for targeting and delivery in the coming years. Medium-term changes envisaged in the sector (related to the potential development of an old-age or disability grant) will also have implications for the assignment of responsibilities, funding flows, and accountability, as will be discussed in the next section. 5.3 Institutional arrangements for policy and oversight 237. Related to these issues of decentralisation is the need to clarify and strengthen the institutional arrangements for policy and oversight. MINALOC has made some very impressive steps forward in recent years in its leadership of the social protection sector and its capacity for technical coordination and oversight. However, as noted in several places above, there are still some gaps in the relationships of accountability which will need to be addressed if the implementation plan is to be successfully put into practice and service delivery to be its most efficient and effective. 238. There is still a need for MINALOC to play a stronger role in oversight of the sector including, in particular, holding its agencies – FARG and RLDSF/VUP – accountable for results, but also in playing a stronger role in policy leadership, and in better coordinating the input of the disparate ‘complementary’ activities into the objectives of the wider social protection sector. 107 Oversight of delivery on the ground: institutional structures for VUP 239. The former will require firstly a clearer assignment of roles and responsibilities for delivery, secondly the appropriate institutional arrangements, and thirdly the appropriate human resources at all levels. Now that funds have been ‘decentralised’ and no longer flow through VUP to the sectors, it is not entirely clear how VUP can be responsible for delivery of services, somewhat undermining its role as an agency. Clarity is urgently needed on the responsibilities and expectations of VUP in its role of programme coordination and monitoring of implementation by the districts, of districts’ responsibilities to report back to VUP, and for VUP to report back to MINALOC. 240. Defining these roles more precisely is of great importance not only for the implementation of current programmes, but also in looking towards the future. Over the medium and long term, as there is both consolidation of existing programmes and the potential for adding new ones (such as the old age grant), the ‘agency’ model of VUP will need to be strengthened, so that rather than VUP being a stand-alone agency delivering only one programme, it develops into a social protection agency delivering a range of programmes. This would ensure that administrative costs are minimised (since each programme does not require entirely separate overhead costs), and would help to ensure overall coherence in programming. 241. Related to this, there are some questions about the lines of accountability and the assignment of functions. For example: is RLDSF as an organisation sufficiently oriented around the needs of delivering social transfers49? Will the current arrangement of shared functions, for example M & E, ensure that the specific needs of VUP are met? Are there really economies of scale in these functions being shared between programmes with such different objectives (VUP vs. RLDSF infrastructure)? 49 For example, some of the current critiques of VUP’s public works components by the recent DFID review (Devereux 2011) relate to the fact that the public works component seems to be focusing more on the production of assets rather than providing social protection. This may be partly related to the fact that VUP is part of RLDSF, whose objectives are more clearly centered on infrastructure development. 108 242. The fact that these questions emerge is entirely a sign of success for the programme and the sector as a whole, and it is not unusual for there to be changes in institutional structures as a sector matures. The point is that the continual adaptation of the original arrangements (for example simply adding new programmes under VUP or RLDSF) may actually not represent the optimal arrangement. What was most appropriate for the last five years may not be the best solution for the next 5, 10, or 20 years. Consideration of the functional responsibilities required for the delivery of social protection and the best institutional structure should therefore be a high priority when feasibility studies are done (for example for the old age grant). Oversight: MINALOC staffing 243. The other side of the coin in terms of oversight of delivery is the ‘demand’ side of accountability. In other words, not only does there need to be a clearer delineation of roles and responsibilities and the appropriate institutional structures to support delivery on the ground, but also MINALOC needs to have the internal capacity to set policy and demand results. 244. The implementation plan envisages that four technical assistance posts will be established including:  Senior Social Protection Adviser.  Social Protection Adviser.  Financial management specialist.  Monitoring and evaluation specialist.  Economist. 245. These will include a combination of national and international recruitment, with the role of the international advisors to also build the capacity of the government staff. Given the likely delays in recruiting long-term posts, development partners will provide support for some short-term TA to be put in place in the interim. 109 246. Filling these posts will be essential to ensure that MINALOC has the capacity to lead future efforts on policy development and coordination, undertake evaluations50 of social protection activities, and ensure high quality reporting to the Joint Sector Reviews. Leadership and coordination of the sector as a whole: MINALOC 247. As part of these policy and oversight functions, MINALOC also needs to work on embedding and internalising the strategy and implementation plan with respect to the clear delineation of the sector (‘core’ versus ‘complementary’ activities). As of the last Joint Sector Review, this clear separation of ‘complementary’ activities has not yet happened, so that the focus remained somewhat scattered. 248. Greater clarity is still needed on what MINALOC’s role is in coordinating the sector with respect to these ‘complementary’ activities. Based on the analysis in the preceding chapters, it is clear that there is still a great deal of overlap and duplication across these complementary activities. While it is not the responsibility of MINALOC for the implementation or even the ultimate design of these areas – as they should fall squarely under other sectors - but there is nevertheless a role to be played in leading the coordination and harmonisation with respect to social protection outcomes. 249. So, for example, this could involve:  Coordinating a review of complementary activities where there are clear overlaps and duplication (for example, in housing, livestock provision, and income generating activities) and leading a policy discussion about how to address potential consolidation and/or harmonisation of these activities. 50 With the strengthening of the evaluation function within MINALOC – which is entirely appropriate in terms of the division of responsibilities – it may also be worthwhile considering the roles and responsibilities for evaluation within other agencies, notably FARG and VUP. It may be that the agencies should focus specifically on monitoring and reporting, with responsibilities for evaluation of higher-level impacts and outcomes resting with the ministry. In other words, it may be good to separate the ‘M’ from the ‘E’. This could be considered as part of a wider functional review of the sector in the medium term. 110  Leading a consultation with stakeholders in particular on the relationship between child protection and social care services and the social protection sector. These areas have been clearly defined as being outside the ‘core’ social protection sector, yet because there is no clear alternative forum for policy coordination and leadership (given that MIGEPROF has not yet taken a strong lead), these areas tend to continue to fall under the social protection banner. It is likely that these areas would be best served by the establishment of a new separate sector with strong links to social protection (in terms of the need for policy coordination) but that is nevertheless separate. 250. These areas have traditionally been lumped together because some of the target populations – as ‘vulnerable groups’ – are the same, however now that the social protection sector has made good progress in clarifying its scope in the strategy – crucially recognising that not all activities related to ‘vulnerable groups’ are social protection, even if they may be complementary - it is important to ensure that this is implemented in practice. Oversight of the sector: the Social Protection Working Group (SPWG) 251. Not only should MINALOC now start to take a stronger leadership role in reinforcing the definition of the sector consistent with the strategy, but also the SP Working Group (SPWG) should ensure that its overall terms of reference, the reporting in the Joint Sector Review, and the interventions of stakeholders are all aligned with this definition of the sector and reinforce a focus on results. Terms of reference of the SPWG 252. As per the Implementation Plan, the structure of the SPWG will evolve so that there is a Leadership Group, and Technical Working Group, as well as various sub- committees focusing on specific areas (for example, finance, monitoring and evaluation, etc.). While this will certainly require a greater level of time commitment on the part of 111 development partners (many of whom may not have the resources to respond fully), it will ensure that the SPWG plays a stronger leadership and oversight role than it is currently playing. The analysis in this assessment has raised a range of issues that could potentially feed into the terms of reference for these different groups that are expected to be elaborated. Joint Sector Review Reporting 253. Currently the reporting format for the JSR is dependent on the template provided by MINECOFIN. In an effort to build on the progress already made in JSR reporting, the following areas could be improved in the short term to enhance the format and substance of reporting required by MINECOFIN:  Clarifying which indicators will be most important in providing the SPWG with adequate information to assess progress, identify areas that need improvement, and set the forward-looking agenda and targets. At present the reporting of progress by programme appears to be somewhat ad hoc, and not necessarily as relevant as it could be.  Ensuring that the reporting provides not only a descriptive assessment but also more detailed analysis of the results and implications for next steps. Examples include providing not just an analysis of expenditure out-turns for the period under review, but also a historical perspective; a clearer assessment of financial out-turns against outputs; and a stronger narrative section in explaining divergences from the budgeted/planned levels. 5.4 Donor coordination: establishment of a SWAp 254. Although the SPWG has been in place for several years, and there has been steady improvement in the levels of harmonisation and coordination, there is a need now to take this one step further through the formal establishment of a social protection sector funding mechanism (SPSFM). This is often referred to as a Sector Wide 112 Approach (SWAp) for social protection, however it is important to bear in mind that a SWAp should be seen more as a process rather than a modality. In many ways, therefore, the sector has already made substantial progress along the SWAp road, but the establishment of a specific funding mechanism in the SPSFM is seen an important next milestone. This section will articulate some of the rationale, the steps required, and some lessons learned from elsewhere that could be helpful for Rwanda. Objectives of a SWAp in social protection 255. So far there seems to be agreement on the following objectives for establishing a SWAp, although different stakeholders probably each have slightly different motivations and preferences:  Making sure all donor activity is ‘on budget’ for greater harmonisation, coordination, and alignment of spending.  Providing a mechanism for funding support that goes beyond VUP to include wider TA/support of sector as a whole, thereby reducing the extent to which much of current TA is ‘ad hoc’ rather than strategic and jointly planned .  Providing a coherent way to articulate sectoral funding needs, financing gaps, and establishing how donor support can meet these in a harmonised and coordinated way. 113 Box 5: Lessons learned from the Health Sector SWAp in Mozambique The Mozambican health sector SWAp has often been cited as a good example of the advantages of working sector-wide. It is therefore a useful case study from which to draw lessons on what worked well, and what hurdles were encountered. The sector wide approach to programming was adopted, with an emphasis on the process through which government and development partners could share a set of common principles, objectives, and working arrangements. The cornerstones of the SWAp were: • A health sector strategic plan, which included a set of indicators to evaluate implementation. • A code of conduct, setting the basic rules of engagement of government and partners. • A set of mechanisms and working arrangements that enabled structured dialogue and consensus building between government and partners, including the SWAp Coordination Committee, the SWAp Forum and various technical working group. • A sector financing framework (integrated into the sector strategic plan and code of conduct) highlighting the expectations of government in relation to aid modalities and financial instruments to be used by development partners; and • A set of review mechanisms. While the SWAp was largely successful in building consensus, aligning donor activities, and increasing the levels of funding flowing through the common donor fund, it faced several challenges, particularly with respect to sustainability and effectiveness. “In essence, the SWAp Forum was beginning to be seen as failing to combine openness and inclusiveness with the required effectiveness to solve problems and provide accountability. This situation was enabling the coexistence of various degrees of commitment to the SWAp principles among both development partners and different MoH departments. The review of the SWAp undertaken in 2004 recommended that while the inclusiveness of the ‘SWAp group’ should be maintained, the commitment to the code of conduct and the objective s of harmonisation should become compulsory and explicit for all its members. The review also recommended that the SWAp Forum should set clear and measurable steps to achieve the objectives of harmonisation, alignment, transparency and accountability.” (Ma rtinez 2006: 5). Source: Javier Martinez (2006) “Implementing a Sector Wide Approach in Health: the Case of Mozambique”, HLSP Institute Technical Approach Paper. 256. Based on the case study of Mozambique, there are therefore some important implications for Rwanda, as summarised in the table below. One of the most important of these is related to the process, and the fact that the basic mechanisms of dialog are well articulated from the outset in the memorandum of understanding. The focus initially should therefore be on the content of the MoU: what do each of the donors 114 need to see? Is there already agreement on objectives, monitoring, and performance benchmarks (and extent to which funding will be withheld if not reached)? If not, a process for dialogue should be managed within the social protection working group. Table 13: Lessons learned in Mozambique and implications for Rwanda Lessons learned in Mozambique Relevance for Rwanda’s social protection sector The sector strategic plan focused on a set of Rwanda’s Implementation Plan probably already broad medium - to long-term goals rather than a addresses many of these problems, as it does clear set of priorities for the short term. This prioritise a set of concrete activities in the short made it difficult to set a clear direction or hold term. government accountable on policy implementation. However, it will be important to ensure that Annual Work Plans provide a realistic set of objectives to be achieved against which progress can be clearly measured. The SWAp emphasised the importance of It will be important to ensure that these enabling mutual accountability between mechanisms of dialogue are well articulated, to government and development partners. ensure that the SWAp has built-in mechanisms However, when progress was not achieved or for dealing with any bumps in the road that may when one side perceived the other side of being arise. at fault, the mechanisms for dialogue within the SWAp architecture became essential. Ultimately the SWAp Forum – an information sharing and consensus building structure – did not have the power to hold its members (donors and government) accountable. Some donors were seen to be participating in the Expectations and how to deal with any non- SWAp but not abiding by its core principles, performance of both donors and government will notably with respect to harmonising aid need to be explicitly addressed in the MoU. instruments. Many of the technical working groups reporting The sub-committees of the TWG that are back to the SWAp forum were not seen to envisaged should be carefully coordinated with produce adequate analysis on which firm the technical staff at MINALOC so that any recommendations could be made. diagnostic analysis can be produced in a timely manner. There was seen to be a lack of systematic follow The ‘leadership’ and ‘technical’ groups will need up by the Ministry (and also by development to ensure that there are clear plans for follow-up partners) of the recommendations made in the embedded in each session. joint annual review reports. This then raised questions about the relevance of the Steering Committee, if reports were endorsed but recommendations were not followed up by implementation plans. 115 6. Conclusions and recommendations 257. What is clear from the assessment of the social protection sector in Rwanda is that very rapid progress has been made in the last 5 years, both in terms of the development and delivery of programmes, but also in the leadership, strategic direction, and coordination of the sector by MINALOC. This assessment therefore provided an opportunity to take stock of the progress so far, and look towards the future as the new strategy will be put in place, with a particular eye on the hurdles that will need to be overcome as the social protection sector moves into its next phase. 258. There are two sets of conclusions that can be reached and recommendations that follow from the preceding analysis. The first pertains to the appropriateness and relevance of the current social protection programmes and strategic vision for the sector. The second is related to both the successes and lessons learned as well as ‘works in progress’ and ‘red flags’ identified in Chapter 4 in terms of specific issues related to design, implementation, financial management, monitoring and evaluation, etc. of different programmes. The third are the higher-level and more strategic conclusions related to the overall consolidation and management of the sector as a whole. 6.1 Appropriateness and relevance of social protection programmes 259. Returning to the analysis of poverty and vulnerability in Chapter 2, we can draw some conclusions about the extent to which current programmes – and the development of the sector as outlined in the Strategy – address the needs of the country. 260. There are clearly gaps in the level of current coverage, with the number of beneficiaries representing only about 4% of the population in total compared to extreme poverty levels of 24%. However, once VUP scales up as planned, that programme alone would be expected to cover around 18% of the population. 261. In terms of the needs of particular groups identified in the poverty and vulnerability analysis, Rwanda’s social protection sector does generally cater for the 116 needs of many of the core groups identified, at least in terms of intended coverage (issues in practice are discussed in the next two sections).  Extremely poor households with no labour capacity are included in VUP direct support. This should include the elderly who are no longer able to support themselves, female-headed households without adequate labour power, and households with persons with disabilities who similarly lack in adequate labour resources.  Extremely poor households with labour capacity (which would presumably include a large percentage of agricultural wage labour households identified as amongst the poorest and the most low-income agricultural households vulnerable to weather-related shocks) are eligible for VUP Public Works. For the agricultural labour households in particular, this approach using public works employment – if sufficient days of employment are provided (see below for related issues in implementation) – should have the additional benefit of reducing the dependency of these households on richer farmers in their villages. This could increase their bargaining power, and also impact on issues related to social exclusion if such alternative employment allows for a greater sense of dignity and self-reliance51.  FARG Direct Support activities are designed to reach orphans, the elderly unable to work, and those survivors living with disabilities who are without other support. 262. Whether these groups are actually reached in practice - and, importantly, whether other categories of the chronically poor and vulnerable are covered – is a matter for further evaluation of the sector. The Social Protection Strategy is explicitly based on a progressive realisation of a social protection floor, which envisages 51 However, this pre-supposes that targeting through Ubudehe does identify agricultural wage laborers as VUP beneficiaries; in practice those who are most severely socially excluded may not be identified by their communities to benefit. This will need further study in the upcoming evaluations planned for the sector. 117 expanding coverage over time. The planned feasibility studies for a potential old age and disability grant should also ensure that these questions about appropriateness and coverage are addressed so that they can inform future directions for the sector. 6.2 Issues related to the effectiveness of specific programmes 263. The successes of VUP are numerous, not least the rapid development and implementation of the programme; the ability to put in place a set of operational procedures and staff training; the consistent and predictable scaling-up of the programme to a new lot of sectors each year; and, perhaps most importantly, the ongoing refinement of the process to increase efficiency and effectiveness. The Ubudehe targeting approach may not be perfect, but it seems to perform at least as well as other poverty-targeted approaches, and most importantly it has a very high degree of buy-in and ownership, so much so that it is implemented even beyond current VUP sectors so that other programmes such as the mutuelle de santé can utilise it. Districts also appear to have a high degree of ownership over social protection activities through VUP, mirroring the political commitment at all levels of government. 264. These are achievements that are especially notable because they are not commonly found in other countries in the region, where donor-funded pilot programmes tend to be small in nature rather than national and government owned; scale-up happens slowly if at all; there is a strong lack of political commitment; implementation is hampered by ineffective targeting design; and activities are dominated by uncoordinated and largely off-budget donor projects. By contrast with this norm, Rwanda’s social protection sector is a clear success story for the region as well as beyond. 265. The flip side of this success and indeed Rwanda’s own ambition s for the sector is therefore precisely the identification of the next round of issues to be resolved, as identified in Chapter 4. 118 266. Issues related to programme design:  The definition of ‘graduation’ is problematic because it relies only on the subjective assessment of household’s status from one year to the next, and the same criteria for eligibility determine ‘graduation’. This latter issue is likely to undermine the ability of the programme to ensure long-term ‘graduation’: if households have their benefits removed as soon as they reach a higher standard of living they are likely to fall back down as soon as the support stops.  Ubudehe targeting may run into problems of effectiveness, especially now that Ubudehe categories are being used as eligibility criteria for more programmes.  The current levels of support provided through public works (in terms of the size of transfers received by households) are very small, and therefore the social protection provided by that component of VUP may in practice be smaller than originally envisaged in the programme design. 267. Issues related to programme implementation:  The lack of a coordinated MIS system hinders the ability to effectively monitor implementation, although this is currently being addressed and a sector MIS system is currently being designed.   The timeliness and predictability of transfers under VUP needs to be addressed. Currently delays are caused by delays in implementing the Targeting Survey as well as lags in districts requesting disbursements.   Currently there is no way to account for unspent balances in VUP financial reporting, reflecting some deficiencies in both financial monitoring reports but also the budgeting process. 119  Practices in VUP public works of including non-eligible households in order to ensure quick implementation of projects needs to be addressed, as this undermines the targeting design of the programme.  As the initial public projects in terracing are exhausted, it will be a challenge for districts to identify projects that can be implemented to the requisite quality standard without significantly reducing the percentage of project expenditure going towards wages (as opposed to other inputs).  There is a need for a more systematic approach across programmes to fill the existing gaps in evaluation and ensure conceptual clarity across programmes when assessing performance and making inferences about appropriate policy responses. 6.3 Higher-level strategic issues for the sector 268. The discussions in Chapters 4 and 5 raised a set of more strategic issues for the sector, many of which are inter-related: Harmonisation and consolidation of the sector:  The Strategy and Implementation Plan have set out the vision for ongoing harmonisation and consolidation of the sector, however this review found that before moving forward this will require further feasibility studies and dialogue on which groups will win and lose from this harmonisation. As the sector stands currently, ‘harmonising’ FARG Direct Support with VUP Direct Support would in practice mean that many current beneficiaries would simply be dropped, even if they are still legitimately needy according to FARG’s criteria. This is likely to contradict the current statutory commitments to genocide survivors as stipulated in the law on FARG. 120  Over the medium term, however, it is likely that there will be potential for a more seamless consolidation of programmes oriented around old age and disability benefits (which would encompass both current FARG and VUP Direct Support beneficiaries). Institutional arrangements and decentralisation:  Related to these issues of eventual consolidation of programmes is the question of the ideal institutional arrangements within Rwanda’s decentralised context. Finding the right balance of central and local engagement in social protection will require some further careful thought, however it is clear that the sector should not become fully devolved as is sometimes envisaged. Social protection is very likely to be best conceived as a shared function, and determining exactly which functions should be given to districts, as opposed to an implementing agency such as RLDSF, will require further study and discussion.  At the central level, it will be important for MINALOC to take an increasing role in the leadership of the sector. This, in turn, will require that staffing for the central policy unit is brought up to the levels indicated in the implementation plan, thereby allowing greater ministry inputs in terms of sector oversight and coordination through the Joint Sector Reviews and, crucially, in matters of public finance management, especially improving the consistency and coherence of the planning and budgeting in the sector. Donor coordination: 269. In many ways, the sector has already made some movement towards a SWAp, in terms of establishing joint donor and government reporting and review, and many of the large donors already operate at least partly through budget support. However, there are significant gains which can be recognized if this process continues. The Social Protection Strategy articulates the objective of moving towards a Sector Wide Approach (SWAp) which would: 121 • Move more donor activity ‘on budget’ for greater harmonisation, coordination, and alignment. • Provide a mechanism for funding support going beyond VUP to include the sector as a whole. • Provide a coherent way to articulate total sectoral funding needs, financing gaps, and how donor support can meet these in a harmonised and coordinated way. 6.4 Recommendations 270. In order to address these issues related to specific programmes as well as the wider strategic issues for the sector, there are a set of recommendations as follows. (1) Ensure MINALOC has the proper human resource base for strengthened leadership and oversight of the sector First and foremost, in terms of process, the priorities are, firstly ensuring that MINALOC human resource capacities are strengthened. This is a pre-condition for the rest of the recommendations that follow: (2) Improve oversight and coordination of the sector through the Joint Sector Review processes Reporting should be improved, so that there is analysis and interpretation of results in implementation, as well as concrete steps for addressing any issues involved. To cement these moves and to continue progressing further, it will be important to ensure that Annual Work Plans provide a realistic set of objectives to be achieved against which progress can be clearly measured. 122 Technical working groups should be tasked with addressing the specific issues outlined in Chapter 4 (the ‘red flags’ and ‘works in progress’) as well as other specific issues that have been identified. These include the development of: • A work plan for the investigation of targeting issues in VUP (including ensuring only eligible beneficiaries participate, assessing the extent to which the Ubudehe system continues to work now that multiple programmes depend on it, etc.). To what extent is the Ubudehe targeting approach appropriate for all programmes? The current emphasis on ‘consolidation’ seems to be based largely around harmonising targeting around Ubudehe criteria. This will reduce some transaction costs, but may also lead to greater targeting errors: current diversity of targeting may ‘diversify’ targeting sufficiently to ensure that those missed out by Ubudehe approach are picked up by other programmes. In terms of diagnostic work that could be pursued towards this aim, this could include both a qualitative and a quantitative element within a targeting review. For the qualitative work, it would be helpful to conduct a small piece of research to understand the community-level dynamics around targeting now that mutuelle subsidisation is also tied to the Ubudehe category. For the quantitative work, the EICV III data – while not likely to have sample sizes large enough for evaluation purposes of VUP – could perhaps be matched with the Ubudehe database (along similar lines to the process used in the matching exercise in Chapter 5), with the permission and support of the National Institute of Statistics. This would allow a more robust evaluation of the Ubudehe targeting against the quantitative measure of consumption poverty, as well as other dimensions of deprivation.  A ‘Graduation Policy’ for VUP, which could build on the findings of the targeting study. 123  A plan to ensure PFM issues in the smooth flow of funds from centre to districts for VUP are carefully monitored by the TWG and addressed as necessary. Other PFM issues to be addressed by the working group on Finance will be the way in which balances are carried forward and repayments on Financial Services component of VUP are budgeted and reported  An understanding on the expectations for the number of days of work (and hence total benefit values) to be provided to public works households, as well as the impact on beneficiaries if the relative costs of other inputs increases as activities other than terracing are undertaken.  Coordination mechanisms for IGA and shelter programmes under the ‘social development’ component of the sector.  A review of the evaluation framework across programmes. Topics that should be addressed by the evaluation framework are already listed in an annex to the Implementation Plan. The evaluation framework could incorporate these topics as well as describe the data/survey requirements (including ideally an exploration of the options for a proper impact evaluation of VUP based on the scale-up plan). (3) Establish a process for the development of deepened joint donor/government coordination Mechanisms of dialogue should be well articulated, to ensure that the SWAp which emerges has built-in mechanisms for dealing with any bumps in the road that may arise. There is a need to be clear from outset about expectations of donors for participation in SWAp, and setting clear milestones for their harmonisation of aid instruments. 124 (4) Lead a process of strategic thinking for the medium- and long-term development of the sector, in line with the vision in the Strategy and Implementation plan. This will include:  In the short term, establishing the potential gains from and feasibility of delivering FARG Direct Support (using the current targeting processes) through RLDSF.  Undertaking the feasibility assessment for the old age grant as envisaged in the work plan for the year, including the micro-simulation of different programme options (looking at impacts and financing requirements52) as well as a functional review. This will serve as the basis for:  Dialogue on government commitments (statutory or otherwise) to different vulnerable groups. The key question to examine is: What is the best way to balance the desire for consolidation of programmes with the desire to continue to reach groups currently covered? It would perhaps be helpful to start with clear statement on commitments (statutory or otherwise), especially with respect to FARG direct support but also disabled former combatants. This will help to clarify for all stakeholders where there is scope for consolidation of benefit packages and where certain existing commitments must continue to be met (even if under more harmonised and coordinated implementation arrangements). 52 Note that it should be possible to incorporate the micro simulation of other options (such as the child grant envisaged in the social protection floor) into the feasibility study on the old age grant, since many different types of programmes can normally be easily modeled together for comparison. 125  The clarification of strategic objectives with respect to decentralisation of social protection sector, leading to the development clearly articulated and consistent function assignments, financial flows and administrative arrangements. (5) Translate this movement towards the implementation of the strategic vision and the increasing coordination and coherence of the sector into improvements in the planning and budget/MTEF processes This will require engagement with MINECOFIN on how best to ensure coordination and coherence within their existing budget process. It is likely that there should be both a short-term approach to simply increasing coordination by MINALOC of its relevant agencies (especially FARG and VUP), as well as a medium-term view of how the budget and MTEF classifications – and processes of accountability – can be better aligned with the definition of the sector as it evolves over the next several years. 126 BIBLIOGRAPHY Published reports Allen, Richard (1999) “’New Public Management’: Pitfalls for Central and Eastern Europe” Public Management Forum, Volume 5, No. 1, January/February 1999. Azfar, Omar, Satu Kähkönen, and Patrick Meagher (2001) “Conditions for Effective Decentralised Governance: A Synthesis of Research Findings” IRIS Center, University of Maryland and the World Bank. Blöchliger, Hansjörg (2009) “The Fiscal Autonomy of Sub-Central Governments: An Update” OECD Network on Fiscal Relations Across Levels of Government. Paris OECD. Blöchliger, Hansjörg (2008) “Market Mechanisms in Public Service Provision” OECD Network on Fiscal Relations Across Levels of Government, Working Paper No. 6. Paris OECD. Blöchliger, Hansjörg and David King (2006) “Fiscal Autonomy of Sub-Central Governments” OECD Network on Fiscal Relations Across Levels of Government, Working Paper No. 2. Paris OECD. Devereux (2011) 2nd Annual Review of DFID Support to the Vision 2020 Umurenge Programme (VUP), Rwanda. Instictiotute of Development Studies Centre for Social Protection. Jütting, Johannes, Elena Corsi and Albrecht Stockmayer (2005) “Decentralisation and Poverty Reduction” OECD Development Centre Policy Insights No. 5. Paris: OECD. Jütting, Johannes, Céline Kauffmann, Ida McDonnell, Holger Osterrieder, Nicholas Pinaud and Lucia Wegner (2004). “Decentralisation and Poverty in Developing Countries: Exploring the Impact” OECD Development Centre Working Paper No. 236. Paris: OECD. ISSA (2011) “Extension of Social Security Coverage to the Informal Sector: A Case of the Rwanda Social Security Board” MINECOFIN (2009) National Social Security Policy. MINALOC (2011) Vision 2020 Umurenge Programme Annual Report 2009/10. MINALOC (2011a) National Social Protection Strategy. MINALOC (2011b) Social Protection Joint Sector Review Summary Report. 127 MINALOC (2010) Assessment of the Decentralisation Process in Rwanda. Final Draft Report, October 2010. MINALOC (2010a) VUP Targeting and Poverty Surveys 2009. MINALOC (2010b) Social Protection Public Expenditure Review. OECD (1997) “Managing Across Levels of Government”. Paris: OECD. Shah, Anwar (2004) “Fiscal Decentralisation in Developing and Transition Economies: Progress, Problems, and the Promise” World Bank Policy Research Working Paper No. 3282, April 2004. Washington, DC: World Bank. Schick, Allen (2008) “Why Most Developing Countries Should Not Try New Zealand Reforms” The World Bank Research Observer, Vol. 13, No 1, February 1998, pp. 123-31. Sutch, Helen (1999) “The Relevance of New Public Management for Transition Countries” Public Management Forum, Volume 5, No. 3, May/June 1999. UNDESA (2000) Decentralisation: Conditions for Success: Lessons from Central and Eastern Europe and the Commonwealth of Independent States. Document ST/ESA/PAD/SER.E/7. New York: United Nations. UNDP 1999 “Decentralisation: A Sampling of Definitions”. Working Paper prepared in connection with the Joint UNDP-Government of Germany evaluation of the UNDP Role in Decentralisation and Local Governance. October 1999. UNDP (2005a) Fiscal Decentralization in Transition Economies: Case Studies from the Balkans and Caucasus. Bratislava: UNDP. UNDP (2005b) UNDP Primer: Fiscal Decentralisation and Poverty Reduction. New York: UNDP. World Bank (2011) Republic of Rwanda Support to the Social Protection Sector Aide Memoire, September 13-23, 2011. World Bank (2004) World Development Report 2004: Making Services Work for Poor People. Washington DC: World Bank. Wylde, Emily, Mickie Schoch, Alta Folscher, Jeremy Holland, and Clare O’Brien (unpublished) Rwanda Integrated Child Poverty and Budget Study. 128 Unpublished programme documentation MINALOC (no date) VUP Proposal for Revised Budget Allocation Guidelines for VUP, memorandum. MINALOC (no date a) VUP Monitoring System Revisions, memorandum. MINALOC (2011) Guidelines for Social Protection Earmarked Transfers. MINAGRI: Guidelines for Girinka eligibility. RDRC: project information on housing component. 129 Annex A Concept Note Social Safety Nets Assessment Republic of Rwanda I. Context 1. Rwanda is a land locked country in sub Saharan Africa (SSA) with one of the highest population density rates of about 36353 people per square kilometer. The country registered impressive growth over the period 2002-2008 where annual economic growth averaged about 6% per annum. During the same period, the good levels of growth did not translate into equally good poverty reduction rates neither was the latter as efficient as Rwanda Government desired to achieve her long term development and poverty reduction objectives. During 2002-2005, poverty reduced by 3.5% points from 60.4% to 56.9%54 with a likely little overall impact since the absolute number of the poor increased by 600,00 people owing partly to an estimated annual population growth rate of 2.7%. 2. Vulnerability55 in Rwanda is estimated to exacerbate poverty with poverty rate among the vulnerable groups recorded at 60%56 thus higher than the national average. The poverty assessment57 further indicates that Inequality between 2002 and 2005/6 also increased from a gini coefficient of 0.47 to 0.5. This is despite the government variant programs and interventions targeted to the different vulnerable groups. 53 Government of Rwanda: Third General Census of Population and Housing, 2002 54 Government of Rwanda: Poverty Analysis for the EDPRS, Final Report, May 2007 55 The 2011 Rwanda National Social Protection Strategy identifies vulnerable groups to include the disabled, elderly, chronically ill and genocide survivors. 56 Government of Rwanda: Poverty Analysis for the EDPRS, Final Report, May 2007. 57 Government of Rwanda: Poverty Analysis for the EDPRS, Final Report, May 2007 130 3. On the other hand, public spending on Social Protection (SP) programs, including those that are not regarded as social safety nets (refer to the paragraph below for a definition of social safety nets) is currently recorded at 8%58 which depicts a 2% points increase over 2006-200959. This shows a relatively high government commitment to social protection programs and to fighting poverty and vulnerability. But the sector remains largely fragmented with planning, budgeting, implementation and monitoring and evaluation of different programs not well harmonized and coordinated to maximize desired impact on poverty reduction. This fragmentation is expected to have adverse effects on effectiveness of the various poverty targeted programs on poverty reduction. 4. The government had recognized the above context and responded by introducing SP as a strategic area in its second generation Poverty Reduction Strategy (PRS) which was termed Economic Development and Poverty Reduction Strategy (EDPRS). SP, represented by the flagship Vision 2020 Umurenge Program (VUP), constitutes one of the 3 core flagships that make up the EDPRS60. This strategic orientation of the sector has already reaped some benefits: (i) a SP Sector Working Group ( SP SWG) that meets regularly brings together development partners (DPs) and government institutions to discuss policy, strategic and technical issues and make recommendations to decision makers to promote a sector wide approach with hope to enhance efficiency and effectiveness of interventions; (ii) a VUP Technical Working Group (VUP TWG) meets regularly to discuss the VUP specific flagship issues and make recommendations; (iii) Joint Sector Reviews are conducted twice every year since 2008 to assess progress in implementation of SP interventions and feed into the EDPRS monitoring; (iv) a Public Expenditure Review (PER) was conducted in 2010 to assess public expenditure related efficiency across various elements of public expenditure; and (v). a National Social Protection Strategy (NSPS) was approved by the cabinet on January 19, 2011 to give high level sector orientation to the various interventions. 58 Government of Rwanda and Oxford Policy Management: Social Protection Public Expenditure Review (PER), October 2010. 59 The first ever SP PER which was carried out in 2006 and had found that public expenditure on SP was 6.1% of government total public expenditure. 60 The other two flagships of the EDPRS are (a) sustainable growth for jobs and exports; (b) governance, including deepening decentralization and enhancing accountability. 131 5. The new NSPS is an integrated strategy. It constitutes four main areas namely 61: (i). Promotion of a comprehensive system of cash transfers based on existing CTs in VUP, FARG and Ex-combatants Program, (ii). Extension of contributory social security and labour standards, (iii). Scaling-up cross-cutting complementary services to implementation of the SP interventions (e,g, gender mainstreaming, savings and cooperatives services, community health e.t.c), and (iv). Integration of interventions that allow timely responsiveness to risks and shocks. The global objective of the NSPS is to build a social protection system that tackles poverty and inequality, enables the poor to move out of poverty, helps reduce vulnerability and protect people from shocks, helps improve health and education among all Rwandans, and contributes to economic growth62. 6. Social Safety Nets (SSN) have been defined variously by different researchers and institutions. Some writers refer to them as welfare programs; others call them social assistance while they are termed as cash transfer programs in other instances. The point of convergence seems to be the fact that SSNs are non contributory transfer programs that are targeted in one way or another to the poor and/or the vulnerable segments of populations63. The SSNs referred to in this concept note are based on this premise and underlying definition derived from a convergence of various definitions that have been given by different researchers and institutions. As much as possible, and within limits of information availability, the assessment will cover both the main formal and informal safety nets in Rwanda. II. Relevance 7. The SSNs constitute a significant part of Rwanda’s SP sector spending. The last SP Public Expenditure Review dated October 2010 estimated that SSNs constituted up to 2.8% of total government spending as of 2009/10 budget. With the existing 61 Government of Rwanda: National Social Protection Strategy, January 2011. 62 Government of Rwanda: National Social Protection Strategy, January 2011. 63 The World Bank: For Protection and Promotion – The Design and Implementation of Effective Social Safety Nets, Grosh, Ninno, Tesliuc and Ouerghi, 2008. 132 fragmentation particularly with regard to programs and interventions that are implemented off government budgeting and planning, it is likely that spending on SSNs could even be more than 2.8%. These safety nets are delivered by different institutions and targeted to different vulnerable groups of the Rwandan population. Relevant institutions include but are not limited to the Ministry of Local Government (department for community development and social affairs), the Ministry of Gender and Family Promotion (MIGEPROF), Vision 2010 Umurenge Program (VUP), Genocide Survivors Fund (FARG), Rwanda Demobilization and reintegration Commission (RDRC), the Ministry of Disaster Management and Refugees Affairs (Refugee Council [RC]), One Cow per poor Family Program (OCFP) under the Ministry of Agriculture (MINAGRI), Civil Society Organizations and Local and International NGOs. Related SSNs are targeted among others to the extremely poor households, historically marginalized groups, the genocide survivors, returnees, demobilized soldiers and ex-combatants. The recently approved National Social Protection Strategy (NSPS) raises a need to conduct feasibility assessments on whether specific SSNs targeted to the elderly, handicapped and vulnerable children should be introduced in addition to the poverty targeted grants. 8. Despite the various improvements over the last 2 years described in section I, the SP sector suffers from weaknesses that also affect the SSNs subsector and these revolve around areas of underdeveloped strategic orientation of the sector; weak coordination mechanisms; a young and developing results orientation; overlaps in some functions and services delivery; weak harmonization of planning (including formulation of budgets and related execution), implementation monitoring, impact evaluation and the demand- side response mechanisms to inform policy refinement and future policy and strategy orientation. The above factors show a strong need for harmonization, alignment and joint work programming by SP policy makers and implementers within government and among development partners. A Joint Implementation Plan for the strategy which was approve recently is being developed and will identify among others, key areas for assessment to generate a sound evidence base. 133 9. This assessment will review knowledge gaps in existing assessments in the sector with particular focus on SSNs. The assessment will review existing SSNs as well as their policy, strategy, design and implementation orientation and generate recommendations for improvements. III. Objectives 10. Against the backdrop of the above rationale, the following will be the key objectives of the proposed SSN assessment. (i) To review the landscape of existing SSN programs, interventions and activities . This includes but will not be limited to taking stock of existing social safety nets in Rwanda: their nature, institutions delivering them; selection criteria, target population; delivery mechanisms, financial costs and sources and any other important characteristics. (ii) To assess institutional arrangements and related programming for existing SSNs at the national and local government levels as well as amongst local and international Non-Governmental Organizations (NGOs). Aspects to be reviewed here will include harmonization arrangements, planning and budgeting, modes of financing, implementation, and monitoring and impact evaluation. (iii) To assess central-local government fiscal relations with regard to financing implementation of SSNs. This objective will include among others assessing past trends in bloc transfers versus earmarked funding and current GoR plans to enhance non- earmarked funding for local governments and implications for financing and implementation of SSNs. (iv) To ascertain synergies, complementarities and overlaps in existing SSNs both at the policy, strategy levels as well as in implementation and delivery mechanisms. 134 (v) To review impact evaluation arrangements including underpinning monitoring and reporting framework for various SSN within and outside the SP sector. This will also involve an assessment of the existing management information system. (vi) To make a set of recommendations that can help the sector overcome existing weaknesses and shortcomings at different levels namely: policy, strategy, implementation, monitoring and evaluation, reporting and service delivery levels. IV. Methodology 11. The assessment will be a desk review of existing policies, strategies, assessments and evaluations and implementation progress reports. The task team will work under the leadership and guidance of the Ministry of Local Government (who are the Lead Government Ministry on SP). Consultations will also be conducted with other relevant government institutions and these include but will not be limited to the Ministry of Finance and Economic Planning, Ministries of Health, Education and Agriculture and Executive Agencies under these ministries which are responsible for coordination, delivery or supervision of delivery of the various SSNs. In addition, the Task Team will also review available documentation at district and sector levels including conducting field visits to selected delivery points to witness systems and realism of the SSNs and inherent characteristics. 12. Although the 2005 House Hold Survey was conducted before a couple of the SSN programs had been started, existing analysis of the survey will be referenced to and more analysis will be done, if needful, to benefit programs that existed and build some baseline for the programs that started after 200564. Consultations will also be held with the Development Partners (DPs) who are active in the SP sector. 64 For example the FARG was started in 1998, VUP was launched in 2007 and its implementation started in 2008, the One Cow per Poor Family Program (Girinka) was started in 2006 while the first phase of the Rwanda Demobilization and Reintegration Commission was started in 1997. 135 13. MINALOC is also in a process of developing an Implementation Plan for the NSPS and this assessment has been identified as one of the priority advisory and analytical activities (AAA). The Implementation Plan will be a key ingredient of the SSN assessment once it will be completed. 14. The focal point for this assessment in the Ministry of Local Government will be the General Directorates of Planning, Monitoring and Evaluation, and of Social Affairs and Community Development in the Ministry of Local Government. The team will also work closely with all other relevant officials in the institutions delivering non contributory transfers. V. Deliverables 15. The main deliverable will be a final report of “An assessment of Social Safety Nets in Rwanda”. Intermediary out puts will include a final concept note and consultants’ terms of reference, an inception report and a draft report following an inception mission and field visits. VI. Collaboration with Local Donor Community 16. This assessment is a collaborative effort with the Government of Rwanda (Ministry of Local Government), the World Bank and Development Partners involved in SP. The work will be coordinated with the planned mapping and rapid assessment of the SP interventions by the International Labour Organization (ILO) and efforts will be made to have joint dissemination of outputs at different stages. 17. Whereas the ILO work will conduct a mapping exercise of all social protection interventions, this assessment will conduct an in-depth analysis of SSNs in Rwanda along the objectives described in section III to generate recommendations at the policy, program and intervention levels. 136 VII. Bank Team composition, Budget and Time line 18. The Bank Task team will constitute of Alex Kamurase (Task Team Leader/Social Protection Specialist), Yasser El-Gammal (Cluster Leader/Lead Social Protection Specialist) and Consultants (1 international consultant, 1 regional consultant and 1 local consultant). 19. The task is one of the main activities to be financed from the Rapid Social Response Multi Donor Trust Fund (RSR MDTF). Related costs will be charged to TF098480 (the Bank Executed [BE] portion of the grant). A summary of the task budget is provided in the table below. Summary task forecast budget in US$ - only tentative No. Item Quantity Unit Total Cost Cost 1. International Consultant 1 x 30 days 750 22,500 2. Regional Consultant 1 x 30 days 600 18,000 3. Local Consultant 1 x 45 days 350 15,570 4. Staff travel 2 trips of 2 weeks each 8,000 16,000 Total 72,250 4. Overheads (transport, workshops, 15% of Total 10,838 focus group meetings, per diem) Grand Total N/A N./A 83,088 20. The Government of Rwanda, Ministry of Local Government has expressed the importance of conducting such an assessment sooner than later and would like to see it delivered earliest possible to both inform the sector generally and specifically the implementation of the NSPS. It is also anticipated that recommendations from the study will be a useful input to the budgetary planning for FY 2011/12 and related sector programming. As such, the team aims at delivering the final product to the government by September 2011 or earliest possible. Activities under this task and estimated time lines are summarized in the table below. 137 Summary activities tentative time lines No Activity Time line Remarks 1. Draft Concept Note (CN) and local donor consultations. March 18, 2011 2. CN review meeting April 15, 2011 WB internal review. TBC 3. Recruitment of consultants April 30, 2011 The timeline 4. Contracts effectiveness April 30, 2011 largely shifted due 5. Inception report June 30, 2011 to the need 6. Draft report August 30, 2011 to wait for 7. Final report September 30, data on 2011 Ubudehe. 8. Validation workshop October 15, 2011 138 Annex B Ubudehe matching exercise methodology note In theory, matching a list of FARG Direct Support beneficiaries with a list of households by Ubudehe status should be a fairly straightforward exercise, since both lists are readily available. In practice, however, it is somewhat more complicated because of inconsistencies across the two lists and cumbersome data formats65. In order to achieve the maximum number of accurate matches, the process involved a matching algorithm based on the ID number. This should in theory be the only data point required to match the two lists, however in practice (a) many FARG DS beneficiaries are minors without ID numbers in the current database (since only adults have IDs); and (b) even among adults there are missing or incorrect ID numbers (due to a large number of typographical errors in recording the IDs in both lists). Where there is no match on ID number, the algorithm then matches on:  First Name  Last Name  District  Sector  Cell The algorithm is able to identify potential matches by name even when there are slight differences in spelling or inconsistencies in given names listed (as is very often the case). Unfortunately, however, the FARG beneficiary list does not contain information on the village of the beneficiary, so for a large number of beneficiaries a unique match was not possible if there are several people with the same name living in the same cell. 65 In particular, the Ubudehe database was provided in a set of separate excel files, with one file per village. The data therefore required manual extraction and transformation from each individual file into a master list, involving 450-600 files per district – a very time-consuming and laborious process. 139 The analysis was undertaken for 3 districts (Musanze, Kicukiro, and Bugesera), which were selected from the 6 districts used as case studies for the Assessment to ensure a range of urban and rural, richer and poorer, and higher and lower performing districts66. The results of the matching are shown in the table below. Overall 39% of beneficiaries on the FARG beneficiary lists for the districts were successfully matched in the Ubudehe database, with a slightly higher percentage matched in Musanze and slightly lower in Bugesera. Table 14: FARG Direct Support beneficiary list matching with Ubudehe database by district Total matched Unmatched Total beneficiaries % matched Musanze 75 91 166 45% Bugesera 298 485 783 38% Kicukiro 246 375 621 40% Total 619 951 1570 39% Age profile of FARG DS beneficiaries The age profile of FARG DS beneficiaries is relevant, remembering that there are essentially three categories of beneficiaries: orphans, disabled adults, and the elderly who are without means of support. We should therefore be able to glean some sense of the current targeting from the age of beneficiaries. Unfortunately the age is only available in the Ubudehe list, so we only have the age for those individuals who have been matched. 66 Ideally the analysis would have included all 6 districts selected as case studies for the Assessment, but the laboriousness of the data transformation process meant that it was only possible to analyse three. 140 As the histograms below illustrate, if we only look at beneficiaries that have been matched successfully, the age profile appears to be centred on those aged 40-60. However, if we include all the beneficiaries with multiple potential matches, we see that the majority of those ‘unmatched’ are children. The data must be taken with caution, since it reflects duplicates in the data, but it nevertheless does confirm that if all beneficiaries were matched the age profile would likely reflect a larger number of children. Table 15: Age distribution of FARG beneficiaries 141 Social Protection & Labor Discussion Paper Series Titles 2012-2014 No. Title 1424 Profiling the Unemployed: A Review of OECD Experiences and Implications for Emerging Economies by Artan Loxha and Matteo Morgandi, August 2014 1423 Any Guarantees? An Analysis of China’s Rural Minimum Living Standard Guarantee Program by Jennifer Golan, Terry Sicular and Nithin Umapathi, August 2014 1422 World Bank Support for Social Safety Nets 2007-2013: A Review of Financing, Knowledge Services and Results by Colin Andrews, Adea Kryeziu and Dahye Seo, June 2014 1421 STEP Skills Measurement Surveys: Innovative Tools for Assessing Skills by Gaëlle Pierre, Maria Laura Sanchez Puerta, Alexandria Valerio and Tania Rajadel, July 2014 1420 Our Daily Bread: What is the Evidence on Comparing Cash versus Food Transfers? by Ugo Gentilini, July 2014 1419 Rwanda: Social Safety Net Assessment by Alex Kamurase, Emily Wylde, Stephen Hitimana and Anka Kitunzi, July 2012 1418 Niger: Food Security and Safety Nets by Jenny C. Aker, Carlo del Ninno, Paul A. Dorosh, Menno Mulder-Sibanda and Setareh Razmara, February 2009 1417 Benin: Les Filets Sociaux au Bénin Outil de Réduction de la Pauvreté par Andrea Borgarello et Damien Mededji, Mai 2011 1416 Madagascar Three Years into the Crisis: An Assessment of Vulnerability and Social Policies and Prospects for the Future by Philippe Auffret, May 2012 1415 Sudan Social Safety Net Assessment by Annika Kjellgren, Christina Jones-Pauly, Hadyiat El-Tayeb Alyn, Endashaw Tadesse and Andrea Vermehren, May 2014 1414 Tanzania Poverty, Growth, and Public Transfers: Options for a National Productive Safety Net Program by W. James Smith, September 2011 1413 Zambia: Using Social Safety Nets to Accelerate Poverty Reduction and Share Prosperity by Cornelia Tesliuc, W. James Smith and Musonda Rosemary Sunkutu, March 2013 1412 Mali Social Safety Nets by Cécile Cherrier, Carlo del Ninno and Setareh Razmara, January 2011 1411 Swaziland: Using Public Transfers to Reduce Extreme Poverty by Lorraine Blank, Emma Mistiaen and Jeanine Braithwaite, November 2012 1410 Togo: Towards a National Social Protection Policy and Strategy by Julie van Domelen, June 2012 1409 Lesotho: A Safety Net to End Extreme Poverty by W. James Smith, Emma Mistiaen, Melis Guven and Morabo Morojele, June 2013 1408 Mozambique Social Protection Assessment: Review of Social Assistance Programs and Social Protection Expenditures by Jose Silveiro Marques, October 2012 1407 Liberia: A Diagnostic of Social Protection by Andrea Borgarello, Laura Figazzolo and Emily Weedon, December 2011 1406 Sierra Leone Social Protection Assessment by José Silvério Marques, John Van Dyck, Suleiman Namara, Rita Costa and Sybil Bailor, June 2013 1405 Botswana Social Protection by Cornelia Tesliuc, José Silvério Marques, Lillian Mookodi, Jeanine Braithwaite, Siddarth Sharma and Dolly Ntseane, December 2013 1404 Cameroon Social Safety Nets by Carlo del Ninno and Kaleb Tamiru, June 2012 1403 Burkina Faso Social Safety Nets by Cécile Cherrier, Carlo del Ninno and Setareh Razmara, January 2011 1402 Social Insurance Reform in Jordan: Awareness and Perceptions of Employment Opportunities for Women by Stefanie Brodmann, Irene Jillson and Nahla Hassan, June 2014 1401 Social Assistance and Labor Market Programs in Latin America: Methodology and Key Findings from the Social Protection Database by Paula Cerutti, Anna Fruttero, Margaret Grosh, Silvana Kostenbaum, Maria Laura Oliveri, Claudia Rodriguez-Alas, Victoria Strokova, June 2014 1308 Youth Employment: A Human Development Agenda for the Next Decade by David Robalino, David Margolis, Friederike Rother, David Newhouse and Mattias Lundberg, June 2013 1307 Eligibility Thresholds for Minimum Living Guarantee Programs: International Practices and Implications for China by Nithin Umapathi, Dewen Wang and Philip O’Keefe, November 2013 1306 Tailoring Social Protection to Small Island Developing States: Lessons Learned from the Caribbean by Asha Williams, Timothy Cheston, Aline Coudouela and Ludovic Subran, August 2013 1305 Improving Payment Mechanisms in Cash-Based Safety Net Programs by Carlo del Ninno, Kalanidhi Subbarao, Annika Kjellgren and Rodrigo Quintana, August 2013 1304 The Nuts and Bolts of Designing and Implementing Training Programs in Developing Countries by Maddalena Honorati and Thomas P. McArdle, June 2013 1303 Designing and Implementing Unemployment Benefit Systems in Middle and Low Income Countries: Key Choices between Insurance and Savings Accounts by David A. Robalino and Michael Weber, May 2013 1302 Entrepreneurship Programs in Developing Countries: A Meta Regression Analysis by Yoonyoung Cho and Maddalena Honorati, April 2013 1301 Skilled Labor Flows: Lessons from the European Union by Martin Kahanec, February 2013 1220 Evaluating the Efficacy of Mass Media and Social Marketing Campaigns in Changing Consumer Financial Behavior by Florentina Mulaj and William Jack, November 2012 1219 Do Social Benefits Respond to Crises? Evidence from Europe & Central Asia During the Global Crisis by Aylin Isik-Dikmelik, November 2012 1218 Building Results Frameworks for Safety Nets Projects by Gloria M. Rubio, October 2012 1217 Pension Coverage in Latin America: Trends and Determinants by Rafael Rofman and Maria Laura Oliveri, June 201 1216 Cash for Work in Sierra Leone: A Case Study on the Design and Implementation of a Safety Net in Response to a Crisis by Colin Andrews, Mirey Ovadiya, Christophe Ribes Ros and Quentin Wodon, November 2012 1215 Public Employment Services, and Activation Policies by Arvo Kuddo, May 2012 1214 Private Pension Systems: Cross-Country Investment Performance by Alberto R. Musalem and Ricardo Pasquini, May 2012 1213 Global Pension Systems and Their Reform: Worldwide Drivers, Trends, and Challenges by Robert Holzmann, May 2012 1211 International Patterns of Pension Provision II: A Worldwide Overview of Facts and Figures by Montserrat Pallares-Miralles, Carolina Romero and Edward Whitehouse, June 2012 1210 Climate-Responsive Social Protection by Anne T. Kuriakose, Rasmus Heltberg, William Wiseman, Cecilia Costella, Rachel Cipryk and Sabine Cornelius, March 2012 1209 Social Protection in Low Income Countries and Fragile Situations: Challenges and Future Directions by Colin Andrews, Maitreyi Das, John Elder, Mirey Ovadiya and Giuseppe Zampaglione, March 2012 1208 World Bank Support for Pensions and Social Security by Mark Dorfman and Robert Palacios, March 2012 1207 Labor Markets in Middle and Low Income Countries: Trends and Implications for Social Protection and Labor Policies by Yoonyoung Cho, David Margolis, David Newhouse and David Robalino, March 2012 1206 Rules, Roles and Controls: Governance in Social Protection with an Application to Social Assistance by Lucy Bassett, Sara Giannozzi, Lucian Pop and Dena Ringold, March 2012 1205 Crisis Response in Social Protection by Federica Marzo and Hideki Mori, March 2012 1204 Improving Access to Jobs and Earnings Opportunities: The Role of Activation and Graduation Policies in Developing Countries by Rita Almeida, Juliana Arbelaez, Maddalena Honorati, Arvo Kuddo, Tanja Lohmann, Mirey Ovadiya, Lucian Pop, Maria Laura Sanchez Puerta and Michael Weber, March 2012 1203 Productive Role of Safety Nets by Harold Alderman and Ruslan Yemtsov, March 2012 1202 Building Social Protection and Labor Systems: Concepts and Operational Implications by David A. Robalino, Laura Rawlings and Ian Walker, March 2012 1201 MicroDeterminants of Informal Employment in the Middle East and North Africa Region by Diego F. Angel-Urdinola and Kimie Tanabe, January 2012 To view Social Protection & Labor Discussion papers published prior to 2012, please visit www.worldbank.org/spl Abstract Rwanda has experienced remarkable reductions in poverty from 59 percent in 2000/01 to 45 percent in 2010/11 (NISR, 2012). Social protection sector has equally evolved recently, both in terms of extending coverage of new programs and developing core functions on policy and strategy. The sector has matured from fragmented and largely off-budget donor programs to an increasingly government-owned and coherent system of interventions. But many Rwandans remain in extreme deprivation and are often stuck in chronic poverty. The sector thus needs to evolve further by deepening systems and reforms to the implementation level, increasing current low coverage of the poor by the core social safety nets programs and adequately responding to the changing needs of poverty and vulnerability. The government remains committed to pro- poor reforms and ensuring inclusive growth is sustained and enhanced. About this series... Social Protection & Labor Discussion Papers are published to communicate the results of The World Bank’s work to the development community with the least possible delay. The typescript manuscript of this paper therefore has not been prepared in accordance with the procedures appropriate to formally edited texts. The findings, interpretations, and conclusions expressed herein are those of the author(s), and do not necessarily reflect the views of the International Bank for Reconstruction and Development / The World Bank and its affiliated organizations, or those of the Executive Directors of The World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The author(s) attest(s) that the paper represents original work. It fully references and describes all relevant prior work on the same subject. For more information, please contact the Social Protection Advisory Service, The World Bank, 1818 H Street, N.W., Room G7-803, Washington, DC 20433 USA. Telephone: (202) 458-5267, Fax: (202) 614-0471, E-mail: socialprotection@worldbank.org or visit us on-line at www.worldbank.org/sp. © 2013 International Bank for Reconstruction and Development / The World Bank